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 please please include enough information in a bug report
9 that someone reading 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.
25 KNOWN PORT-SPECIFIC BUGS
27 The breakpoint-based TRACE facility doesn't work properly in the
28 OpenBSD port of sbcl-0.6.7.
32 (There is also some information on bugs in the manual page and in the
33 TODO file. Eventually more such information may move here.)
35 * The FUNCTION special operator doesn't check properly whether its
36 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
37 instead of failing with an error. (Later attempting to funcall the
38 value does cause an error.)
40 * Failure in initialization files is not handled gracefully -- it's
41 a throw to TOP-LEVEL-CATCHER, which is not caught until we enter
42 TOPLEVEL-REPL. Code should be added to catch such THROWs even when
43 we're not in TOPLEVEL-REPL and do *something* with them (probably
44 complaining about an error outside TOPLEVEL-REPL, perhaps printing
45 a BACKTRACE, then terminating execution of SBCL).
47 * COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
48 * (DEFUN FOO (X) (- 12 X))
50 * (COMPILED-FUNCTION-P #'FOO)
53 * DEFSTRUCT should almost certainly overwrite the old LAYOUT information
54 instead of just punting when a contradictory structure definition
57 * It should cause a STYLE-WARNING, not a full WARNING, when a structure
58 slot default value does not match the declared structure slot type.
59 (The current behavior is consistent with SBCL's behavior elsewhere,
60 and would not be a problem, except that the other behavior is
61 specifically required by the ANSI spec.)
63 * It should cause a STYLE-WARNING, not a WARNING, when the system ignores
64 an FTYPE proclamation for a slot accessor.
66 * Missing ordinary arguments in a macro call aren't reported when the
67 macro lambda list contains &KEY:
68 (DEFMACRO FOO (BAR &KEY) BAR) => FOO
70 Also in DESTRUCTURING-BIND:
71 (DESTRUCTURING-BIND (X Y &REST REST) '(1) (VECTOR X Y REST))
73 Also with &REST lists:
74 (DEFMACRO FOO (BAR &REST REST) BAR) => FOO
77 * Error reporting on various stream-requiring operations is not
78 very good when the stream argument has the wrong type, because
79 the operation tries to fall through to Gray stream code, and then
80 dies because it's undefined. E.g.
81 (PRINT-UNREADABLE-OBJECT (*STANDARD-OUTPUT* 1))
82 gives the error message
83 error in SB-KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
84 The function SB-IMPL::STREAM-WRITE-STRING is undefined.
85 It would be more useful and correct to signal a TYPE-ERROR:
87 (It wouldn't be terribly difficult to write stubs for all the
88 Gray stream functions that the old CMU CL code expects, with
89 each stub just raising the appropriate TYPE-ERROR.)
91 * bogus warnings about undefined functions for magic functions like
92 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
93 like src/code/float.lisp
95 * The "byte compiling top-level form:" output ought to be condensed.
96 Perhaps any number of such consecutive lines ought to turn into a
97 single "byte compiling top-level forms:" line.
99 * The handling of IGNORE declarations on lambda list arguments of DEFMETHOD
100 is at least weird, and in fact seems broken and useless. I should
101 fix up another layer of binding, declared IGNORABLE, for typed
102 lambda list arguments.
104 * Compiling a file containing the erroneous program
108 (DEFSTRUCT (BAR (:INCLUDE FOO))
111 gives only the not-very-useful message
113 (during macroexpansion)
114 Condition PROGRAM-ERROR was signalled.
115 (The specific message which says that the problem was duplicate
116 slot names gets lost.)
118 * The way that the compiler munges types with arguments together
119 with types with no arguments (in e.g. TYPE-EXPAND) leads to
120 weirdness visible to the user:
121 (DEFTYPE FOO () 'FIXNUM)
123 (TYPEP 11 '(FOO)) => T, which seems weird
124 (TYPEP 11 'FIXNUM) => T
125 (TYPEP 11 '(FIXNUM)) signals an error, as it should
126 The situation is complicated by the presence of Common Lisp types
127 like UNSIGNED-BYTE (which can either be used in list form or alone)
128 so I'm not 100% sure that the behavior above is actually illegal.
129 But I'm 90+% sure, and someday perhaps I'll be motivated to look it up..
131 * It would be nice if the
133 (during macroexpansion)
134 said what macroexpansion was at fault, e.g.
136 (during macroexpansion of IN-PACKAGE,
137 during macroexpansion of DEFFOO)
139 * The type system doesn't understand the KEYWORD type very well:
140 (SUBTYPEP 'KEYWORD 'SYMBOL) => NIL, NIL
141 It might be possible to fix this by changing the definition of
142 KEYWORD to (AND SYMBOL (SATISFIES KEYWORDP)), but the type system
143 would need to be a bit smarter about AND types, too:
144 (SUBTYPEP '(AND SYMBOL KEYWORD) 'SYMBOL) => NIL, NIL
145 (The type system does know something about AND types already,
146 (SUBTYPEP '(AND INTEGER FLOAT) 'NUMBER) => T, T
147 (SUBTYPEP '(AND INTEGER FIXNUM) 'NUMBER) =>T, T
148 so likely this is a small patch.)
150 * Floating point infinities are screwed up. [When I was converting CMU CL
151 to SBCL, I was looking for complexity to delete, and I thought it was safe
152 to just delete support for floating point infinities. It wasn't: they're
153 generated by the floating point hardware even when we remove support
154 for them in software. -- WHN] Support for them should be restored.
156 * The ANSI syntax for non-STANDARD method combination types in CLOS is
157 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
158 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
159 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
160 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
161 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
162 the error mesage is not easy to understand:
163 INVALID-METHOD-ERROR was called outside the dynamic scope
164 of a method combination function (inside the body of
165 DEFINE-METHOD-COMBINATION or a method on the generic
166 function COMPUTE-EFFECTIVE-METHOD).
167 It would be better if it were more informative, a la
168 The method combination type for this method (STANDARD) does
169 not match the method combination type for the generic function
171 Also, after you make the mistake of omitting the PROGN qualifier
172 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
174 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
176 INVALID-METHOD-ERROR was called outside the dynamic scope
177 of a method combination function (inside the body of
178 DEFINE-METHOD-COMBINATION or a method on the generic
179 function COMPUTE-EFFECTIVE-METHOD).
180 This is not very helpful..
182 * The message "The top of the stack was encountered." from the debugger
183 is not helpful when I type "FRAME 0" -- I know I'm going to the top
186 * (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
187 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
188 (Also, when this is fixed, we can enable the code in PROCLAIM which
189 checks for incompatible FTYPE redeclarations.)
191 * The ANSI spec says that CONS can be a compound type spec, e.g.
192 (CONS FIXNUM REAL). SBCL doesn't support this.
194 * from Paolo Amoroso on the CMU CL mailing list 27 Feb 2000:
195 I use CMU CL 18b under Linux. When COMPILE-FILE is supplied a physical
196 pathname, the type of the corresponding compiled file is X86F:
197 * (compile-file "/home/paolo/lisp/tools/foo")
198 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:00:46 pm.
199 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
201 Compiling DEFUN SQUARE:
202 Byte Compiling Top-Level Form:
203 /home/paolo/lisp/tools/foo.x86f written.
204 Compilation finished in 0:00:00.
205 #p"/home/paolo/lisp/tools/foo.x86f"
208 But when the function is called with a logical pathname, the file type
210 * (compile-file "tools:foo")
211 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:01:04 pm.
212 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
214 Compiling DEFUN SQUARE:
215 Byte Compiling Top-Level Form:
216 TOOLS:FOO.FASL written.
217 Compilation finished in 0:00:00.
218 #p"/home/paolo/lisp/tools/foo.fasl"
222 * from DTC on the CMU CL mailing list 25 Feb 2000:
223 ;;; Compiler fails when this file is compiled.
225 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
226 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
227 ;;; fails within bind node branch.
229 ;;; Note that if c::*check-consistency* is enabled then an un-reached
230 ;;; entry is also reported.
233 (declare (values nil))
250 (let ((ttt #'(lambda () (go cccc))))
251 (declare (special ttt))
252 (return-from bbbb nil))
255 (return-from bbbb nil))))))
257 * (I *think* this is a bug. It certainly seems like strange behavior. But
258 the ANSI spec is scary, dark, and deep..)
259 (FORMAT NIL "~,1G" 1.4) => "1. "
260 (FORMAT NIL "~3,1G" 1.4) => "1. "
262 * from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
264 (setf (find-class 'ccc1) (find-class 'ccc))
265 (defmethod zut ((c ccc1)) 123)
266 DTC's recommended workaround from the mailing list 3 Mar 2000:
267 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
269 * There's probably a bug in the compiler handling of special variables
270 in closures, inherited from the CMU CL code, as reported on the
271 CMU CL mailing list. There's a patch for this on the CMU CL
273 Message-ID: <38C8E188.A1E38B5E@jeack.com.au>
274 Date: Fri, 10 Mar 2000 22:50:32 +1100
275 From: "Douglas T. Crosher" <dtc@jeack.com.au>
277 * The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
278 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
279 inside "~<..~>" (without the colon modifier on the closing syntax).
280 However, SBCL doesn't do this:
281 * (FORMAT T "~<munge~wegnum~>" 12)
285 * When too many files are opened, OPEN will fail with an
286 uninformative error message
287 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
288 instead of saying that too many files are open.
290 * Right now, when COMPILE-FILE has a read error, it actually pops
291 you into the debugger before giving up on the file. It should
292 instead handle the error, perhaps issuing (and handling)
293 a secondary error "caught ERROR: unrecoverable error during compilation"
294 and then return with FAILURE-P true,
296 * The print system doesn't conform to ANSI
297 "22.1.3.3.1 Package Prefixes for Symbols" for keywords printed when
298 *PACKAGE* is the KEYWORD package.
300 from a message by Ray Toy on CMU CL mailing list Fri, 28 Apr 2000:
302 In a discussion on comp.lang.lisp, the following code was given (by
305 (let ((*package* (find-package :keyword)))
306 (write-to-string object :readably t))
308 If OBJECT is a keyword, CMUCL prints out the keyword, but without a
309 colon. Hence, it's not readable, as requested.
311 I think the following patch will make this work as expected. The
312 patch just basically checks for the keyword package first before
313 checking the current package.
317 --- ../cmucl-18c/src/code/print.lisp Wed Dec 8 14:33:47 1999
318 +++ ../cmucl-18c/new/code/print.lisp Fri Apr 28 09:21:29 2000
319 @@ -605,12 +605,12 @@
320 (let ((package (symbol-package object))
321 (name (symbol-name object)))
323 - ;; If the symbol's home package is the current one, then a
324 - ;; prefix is never necessary.
325 - ((eq package *package*))
326 ;; If the symbol is in the keyword package, output a colon.
327 ((eq package *keyword-package*)
328 (write-char #\: stream))
329 + ;; If the symbol's home package is the current one, then a
330 + ;; prefix is never necessary.
331 + ((eq package *package*))
332 ;; Uninterned symbols print with a leading #:.
334 (when (or *print-gensym* *print-readably*)
336 * from CMU CL mailing list 01 May 2000
338 I realize I can take care of this by doing (proclaim (ignore pcl::.slots1.))
339 but seeing as .slots0. is not-exported, shouldn't it be ignored within the
343 In: DEFMETHOD FOO-BAR-BAZ (RESOURCE-TYPE)
344 (DEFMETHOD FOO-BAR-BAZ
345 ((SELF RESOURCE-TYPE))
346 (SETF (SLOT-VALUE SELF 'NAME) 3))
347 --> BLOCK MACROLET PCL::FAST-LEXICAL-METHOD-FUNCTIONS
348 --> PCL::BIND-FAST-LEXICAL-METHOD-MACROS MACROLET
349 --> PCL::BIND-LEXICAL-METHOD-FUNCTIONS LET PCL::BIND-ARGS LET* PCL::PV-BINDING
350 --> PCL::PV-BINDING1 PCL::PV-ENV LET
352 (LET ((PCL::.SLOTS0. #))
357 Warning: Variable PCL::.SLOTS0. defined but never used.
359 Compilation unit finished.
362 #<Standard-Method FOO-BAR-BAZ (RESOURCE-TYPE) {480918FD}>
364 * reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
366 Also, there is another bug: `array-displacement' should return an array
367 or nil as first value (as per ANSI CL), while CMUCL declares it as
368 returning an array as first value always.
370 * Sometimes (SB-EXT:QUIT) fails with
371 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
372 Process inferior-lisp exited abnormally with code 1
373 I haven't noticed a repeatable case of this yet.
375 * The system accepts DECLAIM in most places where DECLARE would be
376 accepted, without even issuing a warning. ANSI allows this, but since
377 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
378 meaning is rather different, and it's unlikely that the user
379 has a good reason for doing DECLAIM not at top level, it would be
380 good to issue a STYLE-WARNING when this happens. A possible
381 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
382 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
384 * There seems to be some sort of bug in the interaction of the
385 normal compiler, the byte compiler, and type predicates.
386 Compiling and loading this file
387 (IN-PACKAGE :CL-USER)
390 (DECLAIM (FTYPE (FUNCTION (FOO) FOO) FOO-BAR))
391 (DECLAIM (INLINE FOO-BAR))
393 (DECLARE (TYPE FOO FOO))
394 (LET ((RESULT2605 (BLOCK FOO-BAR (PROGN (THE FOO (FOO-A FOO))))))
395 (UNLESS (TYPEP RESULT2605 'FOO)
396 (LOCALLY (ERROR "OOPS")))
397 (THE FOO RESULT2605)))
399 (DEFPARAMETER *FOO* (MAKE-FOO :A (MAKE-FOO)))
400 (UNLESS (EQ *PRINT-LEVEL* 133)
403 (WHEN (TYPEP *X* 'FOO)
406 (PRINT (FOO-BAR *FOO*))
408 in sbcl-0.6.5 (or also in CMU CL 18b for FreeBSD) gives a call
409 to the undefined function SB-C::%INSTANCE-TYPEP. %INSTANCE-TYPEP
410 is not defined as a function because it's supposed to
411 be transformed away. My guess is what's happening is that
412 the mixture of toplevel and non-toplevel stuff and inlining
413 is confusing the system into compiling an %INSTANCE-TYPEP
414 form into byte code, where the DEFTRANSFORM which is supposed
415 to get rid of such forms is not effective.
417 * some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
420 (BLOCK USED-BY-SOME-Y?
423 (UNLESS (REJECTED? Y)
424 (RETURN-FROM USED-BY-SOME-Y? T)))))
425 (DECLARE (INLINE FROB))
430 error in function SB-KERNEL:ASSERT-ERROR:
431 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
433 * The CMU CL reader code takes liberties in binding the standard read table
434 when reading the names of characters. Tim Moore posted a patch to the
435 CMU CL mailing list Mon, 22 May 2000 21:30:41 -0700.
437 * In some cases the compiler believes type declarations on array
438 elements without checking them, e.g.
439 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
442 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
443 (WHEN (CONSP (AREF X 0))
445 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
448 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
449 all cases, e.g. the type assumption *is* checked if the array
450 elements are declared to be of some structure type instead of CONS.
452 * The printer doesn't report closures very well. This is true in
456 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
457 It would be nice to make closures have a settable name slot,
458 and make things like DEFSTRUCT and FLET, which create closures,
459 set helpful values into this slot.
461 * And as long as we're wishing, it would be awfully nice if INSPECT could
462 also report on closures, telling about the values of the bound variables.
464 * as reported by Robert Strandh on the CMU CL mailing list 12 Jun 2000:
466 (defconstant +a-constant+ (make-instance 'a-class))
467 (defconstant +another-constant+ (vector +a-constant+))
469 CMU Common Lisp release x86-linux 2.4.19 8 February 2000 build 456,
472 Send bug reports and questions to your local CMU CL maintainer,
473 or to pvaneynd@debian.org
474 or to cmucl-help@cons.org. (prefered)
475 type (help) for help, (quit) to exit, and (demo) to see the demos
477 Python 1.0, target Intel x86
478 CLOS based on PCL version: September 16 92 PCL (f)
479 * (defclass a-class () ())
480 #<STANDARD-CLASS A-CLASS {48027BD5}>
481 * (compile-file "xx.lisp")
482 Python version 1.0, VM version Intel x86 on 12 JUN 00 08:12:55 am.
484 /home/strandh/Research/Functional/Common-Lisp/CLIM/Development/McCLIM
485 /xx.lisp 12 JUN 00 07:47:14 am
486 Compiling Load Time Value of (PCL::GET-MAKE-INSTANCE-FUNCTION-SYMBOL
488 Byte Compiling Top-Level Form:
489 Error in function C::DUMP-STRUCTURE: Attempt to dump invalid
491 #<A-CLASS {4803A5B5}>
494 * The compiler assumes that any time a function of declared FTYPE
495 doesn't signal an error, its arguments were of the declared type.
496 E.g. compiling and loading
497 (DECLAIM (OPTIMIZE (SAFETY 3)))
498 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
499 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
501 (COND ((> (FACTORIAL X) 1.0E6)
502 (FORMAT T "too big~%"))
504 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
506 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
509 will cause the INTEGERP case to be selected, giving bogus output a la
511 This violates the "declarations are assertions" principle.
512 According to the ANSI spec, in the section "System Class FUNCTION",
513 this is a case of "lying to the compiler", but the lying is done
514 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
515 not by the unexpectedly general definition of FACTORIAL. In any case,
516 "declarations are assertions" means that lying to the compiler should
517 cause an error to be signalled, and should not cause a bogus
518 result to be returned. Thus, the compiler should not assume
519 that arbitrary functions check their argument types. (It might
520 make sense to add another flag (CHECKED?) to DEFKNOWN to
521 identify functions which *do* check their argument types.)
523 * As pointed out by Martin Cracauer on the CMU CL mailing list
524 13 Jun 2000, the :FILE-LENGTH operation for
525 FD-STREAM-MISC-ROUTINE is broken for large files: it says
526 (THE INDEX SIZE) even though SIZE can be larger than INDEX.
528 * In SBCL 0.6.5 (and CMU CL 18b) compiling and loading
529 (in-package :cl-user)
530 (declaim (optimize (safety 3)
532 (compilation-speed 2)
535 #+nil (sb-ext:inhibit-warnings 2)))
536 (declaim (ftype (function * (values)) emptyvalues))
537 (defun emptyvalues (&rest rest) (declare (ignore rest)) (values))
539 (defgeneric assertoid ((x t)))
540 (defmethod assertoid ((x t)) "just a placeholder")
542 (declare (type hash-table ht))
548 (assertoid (hash-table-count ht)))))))
549 (unless (typep res 'foo)
551 (common-lisp-user::bad-result-from-assertive-typed-fun
555 (bar (make-hash-table))
557 Error in KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
558 the function C::%INSTANCE-TYPEP is undefined.
559 %INSTANCE-TYPEP is always supposed to be IR1-transformed away, but for
560 some reason -- the (VALUES) return value declaration? -- the optimizer is
561 confused and compiles a full call to %INSTANCE-TYPEP (which doesn't exist
562 as a function) instead.
564 * DEFMETHOD doesn't check the syntax of &REST argument lists properly,
565 accepting &REST even when it's not followed by an argument name:
566 (DEFMETHOD FOO ((X T) &REST) NIL)
568 * On the CMU CL mailing list 26 June 2000, Douglas Crosher wrote
570 Hannu Rummukainen wrote:
572 > There's something weird going on with the compilation of the attached
573 > code. Compiling and loading the file in a fresh lisp, then invoking
575 Thanks for the bug report, nice to have this one fixed. It was a bug
576 in the x86 backend, the < VOP. A fix has been committed to the main
577 source, see the file compiler/x86/float.lisp.
579 Probably the same bug exists in SBCL.
581 * TYPEP treats the result of UPGRADED-ARRAY-ELEMENT-TYPE as gospel,
582 so that (TYPEP (MAKE-ARRAY 3) '(VECTOR SOMETHING-NOT-DEFINED-YET))
583 returns (VALUES T T). Probably it should be an error instead,
584 complaining that the type SOMETHING-NOT-DEFINED-YET is not defined.
586 * TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
587 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
589 (DECLARE (TYPE INDEXOID X))
590 (THE (VALUES INDEXOID)
592 where the implementation of the type check in function FOO
593 includes a full call to %TYPEP. There are also some fundamental problems
594 with the interpretation of VALUES types (inherited from CMU CL, and
595 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
596 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
598 * The definitions of SIGCONTEXT-FLOAT-REGISTER and
599 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
600 supported on FreeBSD because the floating point state is not saved,
601 but at least as of FreeBSD 4.0, the floating point state *is* saved,
602 so they could be supported after all. Very likely
603 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
605 * (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
606 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same issue
607 with 'union, 'and, 'or etc. So even though according to the ANSI spec,
608 bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL (and now
609 SBCL) interpret them as legal types.
611 * ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
612 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
613 list, and it is probably possible to use substantially the same
614 patches to add it to SBCL.
616 * a slew of floating-point-related errors reported by Peter Van Eynde
618 * (SQRT -9.0) fails, because SB-KERNEL::COMPLEX-SQRT is undefined.
619 Similarly, COMPLEX-ASIN, COMPLEX-ACOS, COMPLEX-ACOSH, and others
621 * SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
622 should probably be 1.4012985e-45. In SBCL,
623 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
624 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
625 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
626 and LEAST-NEGATIVE-LONG-FLOAT.
627 * Many expressions generate floating infinity:
632 PVE's regression tests want them to raise errors. SBCL
633 generates the infinities instead, which may or may not be
634 conforming behavior, but then blow it by being unable to
635 output the infinities, since support for infinities is generally
636 broken, and in particular SB-IMPL::OUTPUT-FLOAT-INFINITY is
638 * (in section12.erg) various forms a la
639 (FLOAT 1 DOUBLE-FLOAT-EPSILON) don't give the right behavior.
641 * type safety errors reported by Peter Van Eynde July 25, 2000:
642 * (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
644 In general lengths of array type specifications aren't
645 checked by COERCE, so it fails when the spec is
646 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
647 * CONCATENATE has the same problem of not checking the length
648 of specified output array types. MAKE-SEQUENCE and MAP and
649 MERGE also have the same problem.
650 * (COERCE 'AND 'FUNCTION) returns something related to
651 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
652 * ELT signals SIMPLE-ERROR if its index argument
653 isn't a valid index for its sequence argument, but should
654 signal TYPE-ERROR instead.
655 * FILE-LENGTH is supposed to signal a type error when its
656 argument is not a stream associated with a file, but doesn't.
657 * (FLOAT-RADIX 2/3) should signal an error instead of
659 * (LOAD "*.lsp") should signal FILE-ERROR.
660 * (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
661 should signal TYPE-ERROR.
662 * MAKE-TWO-WAY-STREAM doesn't check that its arguments can
663 be used for input and output as needed. It should fail with
664 TYPE-ERROR when handed e.g. the results of MAKE-STRING-INPUT-STREAM
665 or MAKE-STRING-OUTPUT-STREAM in the inappropriate positions,
667 * (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
669 should probably signal an error instead of making a pathname with
671 * READ-BYTE is supposed to signal TYPE-ERROR when its argument is
672 not a binary input stream, but instead cheerfully reads from
673 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
675 * DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
676 * (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and doesn't.
677 * (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
678 signal a PROGRAM-ERROR, and doesn't.
679 * (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
680 and other DEFCLASS forms with duplicate specifications in their
681 slots, should signal a PROGRAM-ERROR, and doesn't.
682 * (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
683 causes a COMPILER-ERROR.
685 * SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
686 * (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
687 PROGRAM-ERROR, but SBCL accepts it instead.
688 * SYMBOL-MACROLET should refuse to bind something which is
689 declared as a global variable, signalling PROGRAM-ERROR.
690 * SYMBOL-MACROLET should signal PROGRAM-ERROR if something
691 it binds is declared SPECIAL inside.
693 * LOOP bugs reported by Peter Van Eynde July 25, 2000:
694 * (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
695 the definition of WITH clauses given in the ANSI spec, but
696 compiles and runs happily in SBCL.
697 * a messy one involving package iteration:
698 interpreted Form: (LET ((PACKAGE (MAKE-PACKAGE "LOOP-TEST"))) (INTERN "blah" PACKAGE) (LET ((BLAH2 (INTERN "blah2" PACKAGE))) (EXPORT BLAH2 PACKAGE)) (LIST (SORT (LOOP FOR SYM BEING EACH PRESENT-SYMBOL OF PACKAGE FOR SYM-NAME = (SYMBOL-NAME SYM) COLLECT SYM-NAME) (FUNCTION STRING<)) (SORT (LOOP FOR SYM BEING EACH EXTERNAL-SYMBOL OF PACKAGE FOR SYM-NAME = (SYMBOL-NAME SYM) COLLECT SYM-NAME) (FUNCTION STRING<))))
699 Should be: (("blah" "blah2") ("blah2"))
700 SBCL: (("blah") ("blah2"))
701 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
702 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
704 * type system errors reported by Peter Van Eynde July 25, 2000:
705 * (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
706 but should be (VALUES T T) instead.
707 * (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
708 but should be (VALUES T T) instead.
709 * (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
710 * In general, the system doesn't like '(INTEGER (0) (0)) -- it
711 blows up at the level of SPECIFIER-TYPE with
712 "Lower bound (0) is greater than upper bound (0)." Probably
713 SPECIFIER-TYPE should return NIL instead.
714 * (TYPEP 0 '(COMPLEX (EQL 0)) fails with
715 "Component type for Complex is not numeric: (EQL 0)."
716 This might be easy to fix; the type system already knows
717 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
718 * The type system doesn't know about the condition system,
719 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
720 * The type system isn't all that smart about relationships
721 between hairy types, as shown in the type.erg test results,
722 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
724 * miscellaneous errors reported by Peter Van Eynde July 25, 2000:
726 (DEFGENERIC FOO02 (X))
727 (DEFMETHOD FOO02 ((X NUMBER)) T)
728 (LET ((M (FIND-METHOD (FUNCTION FOO02)
730 (LIST (FIND-CLASS (QUOTE NUMBER))))))
731 (REMOVE-METHOD (FUNCTION FOO02) M)
732 (DEFGENERIC FOO03 (X))
733 (ADD-METHOD (FUNCTION FOO03) M)))
734 should give an error, but SBCL allows it.
735 * READ should probably return READER-ERROR, not the bare
736 arithmetic error, when input a la "1/0" or "1e1000" causes
738 * There are several metaobject protocol "errors". (In order to fix
739 them, we might need to document exactly what metaobject
740 protocol specification we're following -- the current code is
741 just inherited from PCL.)
742 * (BUTLAST NIL) should return NIL. (This appears to be a compiler
743 bug, since the definition of BUTLAST, when interpreted, does
744 give (BUTLAST NIL)=>NIL.)
746 * another error from Peter Van Eynde 5 September 2000:
747 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
748 PVE submitted a patch to deal with this bug, but it exposes other
749 comparably serious bugs, so I didn't apply it. It looks as though
750 the FORMAT code needs a fair amount of rewriting in order to comply
751 with the various details of the ANSI spec.
753 * The bug discussed on the cmucl-imp@cons.org mailing list ca. 5 September,
754 simplified by Douglas Crosher down to
771 causes the same problem on SBCL: compiling it fails with
772 :LET fell through ECASE expression.
773 Very likely the patch discussed there is appropriate for SBCL
774 as well, but I don't understand it, so I didn't apply it.
776 * The implementation of #'+ returns its single argument without
777 type checking, e.g. (+ "illegal") => "illegal".
779 * In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
780 because it's defined with the DEFMACRO-MUNDANELY macro and something
781 is wrong with doc string setting in that macro.
783 * Attempting to use COMPILE on something defined by DEFMACRO fails:
784 (DEFMACRO FOO (X) (CONS X X))
786 Error in function C::GET-LAMBDA-TO-COMPILE:
787 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
789 * In sbcl-0.6.7, the compiler accepted a bogus declaration
790 (TYPE INDEX LENGTH) in the definition of BUTLAST, and then died
791 with infinite regress of errors when the BUTLAST function was
792 executed with a LIST=NIL which would cause LENGTH to be -1.
793 I fixed the bogus declaration, but I should come back and see
794 whether the system's inability to recover from the bogus declaration
795 (by signalling a TYPE-ERROR and dropping into the debugger) was
796 a compiler problem which remains to be fixed, or one of the
797 unrelated infinite-regress-errors problems, many related to
798 revised signal handling, which were fixed around the same time.
800 * Even when FINISH-OUTPUT is called, the system doesn't in general
801 flush the last line of output unless it's terminated by a newline.
802 (This is particularly annoying because several Lisp functions like
803 PRINT *precede* their output with a newline, instead of following