| blob | 8867979e9585674200e2a45630ba4c0cfff0071f |
1 ;;;; This software is part of the SBCL system. See the README file for
2 ;;;; more information.
4 ;;;; This software is derived from software originally released by Xerox
5 ;;;; Corporation. Copyright and release statements follow. Later modifications
6 ;;;; to the software are in the public domain and are provided with
7 ;;;; absolutely no warranty. See the COPYING and CREDITS files for more
8 ;;;; information.
10 ;;;; copyright information from original PCL sources:
11 ;;;;
12 ;;;; Copyright (c) 1985, 1986, 1987, 1988, 1989, 1990 Xerox Corporation.
13 ;;;; All rights reserved.
14 ;;;;
15 ;;;; Use and copying of this software and preparation of derivative works based
16 ;;;; upon this software are permitted. Any distribution of this software or
17 ;;;; derivative works must comply with all applicable United States export
18 ;;;; control laws.
19 ;;;;
20 ;;;; This software is made available AS IS, and Xerox Corporation makes no
21 ;;;; warranty about the software, its performance or its conformity to any
22 ;;;; specification.
25 \f
26 #|
28 The CommonLoops evaluator is meta-circular.
30 Most of the code in PCL is methods on generic functions, including
31 most of the code that actually implements generic functions and method
32 lookup.
34 So, we have a classic bootstrapping problem. The solution to this is
35 to first get a cheap implementation of generic functions running,
36 these are called early generic functions. These early generic
37 functions and the corresponding early methods and early method lookup
38 are used to get enough of the system running that it is possible to
39 create real generic functions and methods and implement real method
41 !FIX-EARLY-GENERIC-FUNCTIONS is called to convert all the early generic
42 functions to real generic functions.
44 The cheap generic functions are built using the same
45 FUNCALLABLE-INSTANCE objects that real generic functions are made out of.
46 This means that as PCL is being bootstrapped, the cheap generic
47 function objects which are being created are the same objects which
48 will later be real generic functions. This is good because:
50 - we can keep pointers to the cheap generic function objects
51 during booting because those pointers will still point to
52 the right object after the generic functions are all fixed up.
54 This file defines the DEFMETHOD macro and the mechanism used to expand
55 it. This includes the mechanism for processing the body of a method.
56 DEFMETHOD basically expands into a call to LOAD-DEFMETHOD, which
57 basically calls ADD-METHOD to add the method to the generic function.
58 These expansions can be loaded either during bootstrapping or when PCL
59 is fully up and running.
61 An important effect of this arrangement is it means we can compile
62 files with DEFMETHOD forms in them in a completely running PCL, but
63 then load those files back in during bootstrapping. This makes
64 development easier. It also means there is only one set of code for
65 processing DEFMETHOD. Bootstrapping works by being sure to have
66 LOAD-METHOD be careful to call only primitives which work during
67 bootstrapping.
69 |#
72 ensure-generic-function-using-class
73 add-method remove-method))
79 ;;; For each of the early functions, arrange to have it point to its
80 ;;; early definition. Do this in a way that makes sure that if we
81 ;;; redefine one of the early definitions the redefinition will take
82 ;;; effect. This makes development easier.
89 name))))
91 ;;; *!GENERIC-FUNCTION-FIXUPS* is used by !FIX-EARLY-GENERIC-FUNCTIONS
92 ;;; to convert the few functions in the bootstrap which are supposed
93 ;;; to be generic functions but can't be early on.
94 ;;;
95 ;;; each entry is a list of the form
96 ;;;
97 ;;; (GENERIC-FUNCTION-NAME METHOD-COMBINATION-NAME METHODS)
98 ;;;
99 ;;; where methods is a list of lists of the form
100 ;;;
101 ;;; (LAMBDA-LIST SPECIALIZERS QUALIFIERS METHOD-BODY-FUNCTION-NAME)
102 ;;;
103 ;;;,where SPECIALIZERS is a list of class names.
106 standard
109 ()
110 real-add-method))
113 standard
116 ()
117 real-remove-method))
120 standard
123 ()
124 real-get-method))
127 standard
129 &key generic-function-class environment
132 ()
133 real-ensure-gf-using-class--generic-function)
135 &key generic-function-class environment
138 ()
139 real-ensure-gf-using-class--null))
142 standard
145 ()
146 real-make-method-lambda))
149 standard
151 lambda-expression environment)
153 ()
154 real-make-method-lambda-using-specializers))
157 standard
160 ()
161 real-make-method-specializers-form))
164 or
183 standard
186 ()
190 ()
194 ()
198 ()
202 ()
206 ()
210 ()
214 ()
218 standard
221 ()
222 real-parse-specializer-using-class))
225 standard
228 ()
229 real-unparse-specializer-using-class))
232 standard
234 lambda-expression
235 lambda-list environment)
237 ()
238 real-make-method-initargs-form))
241 standard
244 ()
245 standard-compute-effective-method)
248 ()
249 short-compute-effective-method))))
250 \f
276 DEFGENERIC: ~S~:@>"
277 spec))
279 ;; FIXME: this list is slightly weird.
280 ;; ANSI (on the DEFGENERIC page) in one
281 ;; place allows only OPTIMIZE; in
282 ;; another place gives this list of
283 ;; disallowed declaration specifiers.
284 ;; This seems to be the only place where
285 ;; the FUNCTION declaration is
286 ;; mentioned; TYPE seems to be missing.
287 ;; Very strange. -- CSR, 2002-10-21
289 inline notinline special))
291 not allowed inside DEFGENERIC."
292 spec))
297 spec))))
298 (:method-combination
303 ~S"
307 (:argument-precedence-order
315 :ARGUMENT-PRECEDENCE-ORDER clause: ~
316 ~S and ~S"
317 required supplied))
326 (:method
329 ;; ANSI requires that unsupported things must get a
330 ;; PROGRAM-ERROR.
350 ;; Hmm. This is similar to BECOME-DEFINED-FUN-NAME
351 ;; except that it doesn't clear an :ASSUMED-TYPE. Should it?
359 :name fun-name
367 fun-name
368 :lambda-list lambda-list
369 'source source-location
370 initargs))
374 ()
379 :format-control format-control
388 lambda-list
393 :context context
395 ;; PARSE-LAMBDA-LIST validates the skeleton, so just check for
396 ;; incorrect use of defaults.
399 "~@<Invalid ~A argument specifier ~S ~_in ~A ~:S~:>"
400 kind arg context lambda-list))
407 ;; no defaults or supplied-p vars allowed for &OPTIONAL or &KEY
414 ~/sb-ext:print-symbol-with-prefix/, ~S, is not a lambda ~
415 form.~@:>"
416 context method-lambda))
417 method-lambda)
418 \f
421 ;;; As per CLHS -
422 ;;; "defmethod is not required to perform any compile-time side effects."
423 ;;; and we don't do much other than to make the function be defined,
424 ;;; which means that checking of callers' arglists can only occur after called
425 ;;; methods are actually loaded.
431 ;; :compile-toplevel is needed for subsequent forms
432 ;; :execute is needed for references to itself inside the body
434 ;; KLUDGE: this double expansion is quite a monumental
435 ;; workaround: it comes about because of a fantastic interaction
436 ;; between the processing rules of CLHS 3.2.3.1 and the
437 ;; bizarreness of MAKE-METHOD-LAMBDA.
438 ;;
439 ;; MAKE-METHOD-LAMBDA can be called by the user, and if the
440 ;; lambda itself doesn't refer to outside bindings the return
441 ;; value must be compileable in the null lexical environment.
442 ;; However, the function must also refer somehow to the
443 ;; associated method object, so that it can call NO-NEXT-METHOD
444 ;; with the appropriate arguments if there is no next method --
445 ;; but when the function is generated, the method object doesn't
446 ;; exist yet.
447 ;;
448 ;; In order to resolve this issue, we insert a literal cons cell
449 ;; into the body of the method lambda, return the same cons cell
450 ;; as part of the second (initargs) return value of
451 ;; MAKE-METHOD-LAMBDA, and a method on INITIALIZE-INSTANCE fills
452 ;; in the cell when the method is created. However, this
453 ;; strategy depends on having a fresh cons cell for every method
454 ;; lambda, which (without the workaround below) is skewered by
455 ;; the processing in CLHS 3.2.3.1, which permits implementations
456 ;; to macroexpand the bodies of EVAL-WHEN forms with both
457 ;; :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL only once. The
458 ;; expansion below forces the double expansion in those cases,
459 ;; while expanding only once in the common case.
470 lambda-list body env)))
486 ;;; Take a name which is either a generic function name or a list specifying
487 ;;; a SETF generic function (like: (SETF <generic-function-name>)). Return
488 ;;; the prototype instance of the method-class for that generic function.
489 ;;;
490 ;;; If there is no generic function by that name, this returns the
491 ;;; default value, the prototype instance of the class
492 ;;; STANDARD-METHOD. This default value is also returned if the spec
493 ;;; names an ordinary function or even a macro. In effect, this leaves
494 ;;; the signalling of the appropriate error until load time.
495 ;;;
496 ;;; Note: During bootstrapping, this function is allowed to return NIL.
503 ; error at load time.
508 \f
509 ;;; These are used to communicate the method name and lambda-list to
510 ;;; MAKE-METHOD-LAMBDA-INTERNAL.
515 body env)
517 ;; don't care to figure out what parts of PCL would have
518 ;; to change to accept that, so coerce.
527 proto-gf proto-method qualifiers specializers method-lambda env))
530 proto-gf proto-method method-function-lambda initargs env))
533 proto-gf proto-method specializers env)))
538 specializers lambda-list)
539 ;; Note: We could DECLAIM the ftype of the generic function here,
540 ;; since ANSI specifies that we create it if it does not
541 ;; exist. However, I chose not to, because I think it's more
542 ;; useful to support a style of programming where every generic
543 ;; function has an explicit DEFGENERIC and any typos in DEFMETHODs
544 ;; are warned about. Otherwise
545 ;;
546 ;; (DEFGENERIC FOO-BAR-BLETCH (X))
547 ;; (DEFMETHOD FOO-BAR-BLETCH ((X HASH-TABLE)) ..)
548 ;; (DEFMETHOD FOO-BRA-BLETCH ((X SIMPLE-VECTOR)) ..)
549 ;; (DEFMETHOD FOO-BAR-BLETCH ((X VECTOR)) ..)
550 ;; (DEFMETHOD FOO-BAR-BLETCH ((X ARRAY)) ..)
551 ;; (DEFMETHOD FOO-BAR-BLETCH ((X LIST)) ..)
552 ;;
553 ;; compiles without raising an error and runs without raising an
554 ;; error (since SIMPLE-VECTOR cases fall through to VECTOR) but
555 ;; still doesn't do what was intended. I hate that kind of bug
556 ;; (code which silently gives the wrong answer), so we don't do a
557 ;; DECLAIM here. -- WHN 20000229
563 initargs-form)))
570 method-class-name initargs-form)
573 fn-lambda)
586 specializers)
597 ;; CONSTANT-FORM-VALUE? What I
598 ;; kind of want to know, though,
599 ;; is what happens if we don't do
600 ;; this for some slow-method
601 ;; function because of a hairy
602 ;; lexenv -- is the only bad
603 ;; effect that the method
604 ;; function ends up unnamed? If
605 ;; so, couldn't we arrange to
606 ;; name it later?
608 specl))
609 specializers))
617 name qualifiers `',specls
618 unspecialized-lambda-list method-class-name
620 #',mname
623 name qualifiers
624 specializers
625 #+nil
631 specializers))
632 unspecialized-lambda-list
633 method-class-name
634 initargs-form))))
638 method-class-name initargs-form)
640 ',method-class-name
641 ',name
642 ',qualifiers
643 ,specializers-form
644 ',unspecialized-lambda-list
645 ,initargs-form
654 proto-gf proto-method method-function-lambda initargs env)))
657 method-lambda initargs env)
669 (check-method-lambda method-lambda 'make-method-lambda) ; TODO remove check in make-method-lambda
670 ;; Default behavior: delegate to MAKE-METHOD-LAMBDA.
682 ;;; When bootstrapping PCL MAKE-METHOD-LAMBDA starts out as a regular
683 ;;; function: REAL-MAKE-METHOD-LAMBDA set to the fdefinition of
684 ;;; MAKE-METHOD-LAMBDA. Once generic functions are born,
685 ;;; REAL-MAKE-METHOD-LAMBDA is used to implement the default method.
686 ;;; MAKE-METHOD-LAMBDA-INTERNAL is split out into a separate function
687 ;;; so that changing it in a live image is easy, and changes actually
688 ;;; take effect.
702 ;;; A helper function for creating Python-friendly type declarations
703 ;;; in DEFMETHOD forms.
704 ;;;
705 ;;; This function operates on
706 ;;; * non-parsed specializers, i.e. class names and extended
707 ;;; specializer syntaxes
708 ;;; * parsed specializers, i.e. CLASSes, EQL-SPECIALIZERs,
709 ;;; CLASS-EQ-SPECIALIZERs and generic SPECIALIZERs
710 ;;;
711 ;;; We're too lazy to cons up a new environment for this, so we just
712 ;;; pass in the list of locally declared specials in addition to the
713 ;;; old environment.
723 ;; KLUDGE: PCL, in its wisdom, sometimes calls methods with
724 ;; types which don't match their specializers. (Specifically,
725 ;; it calls ENSURE-CLASS-USING-CLASS (T NULL) with a non-NULL
726 ;; second argument.) Hopefully it only does this kind of
727 ;; weirdness when bootstrapping.. -- WHN 20000610
730 ;; Independent of SPECIALIZER, bail out if the PARAMETER is
731 ;; known to be a special variable. Our rebinding magic for SETQ
732 ;; cases doesn't work right there as SET, (SETF SYMBOL-VALUE),
733 ;; etc. make things undecidable.
737 ;; Bail out on SLOT-OBJECT special case.
738 ;;
739 ;; KLUDGE: For some low-level implementation classes, perhaps
740 ;; because of some problems related to the incomplete
741 ;; integration of PCL into SBCL's type system, some specializer
742 ;; classes can't be declared as argument types. E.g.
743 ;; (DEFMETHOD FOO ((X SLOT-OBJECT))
744 ;; (DECLARE (TYPE SLOT-OBJECT X))
745 ;; ..)
746 ;; loses when
747 ;; (DEFSTRUCT BAR A B)
748 ;; (FOO (MAKE-BAR))
749 ;; perhaps because of the way that STRUCTURE-OBJECT inherits
750 ;; both from SLOT-OBJECT and from SB-KERNEL:INSTANCE. In an
751 ;; effort to sweep such problems under the rug, we exclude these
752 ;; problem cases by blacklisting them here. -- WHN 2001-01-19
756 ;; Bail out on unparsed EQL-specializers.
757 ;;
758 ;; KLUDGE: ANSI, in its wisdom, says that EQL-SPECIALIZER-FORMs
759 ;; in EQL specializers are evaluated at DEFMETHOD expansion
760 ;; time. Thus, although one might think that in
761 ;; (DEFMETHOD FOO ((X PACKAGE)
762 ;; (Y (EQL 12))
763 ;; ..))
764 ;; the PACKAGE and (EQL 12) forms are both parallel type names,
765 ;; they're not, as is made clear when you do
766 ;; (DEFMETHOD FOO ((X PACKAGE)
767 ;; (Y (EQL 'BAR)))
768 ;; ..)
769 ;; where Y needs to be a symbol named "BAR", not some cons made
770 ;; by (CONS 'QUOTE 'BAR). I.e. when the EQL-SPECIALIZER-FORM is
771 ;; (EQL 'X), it requires an argument to be of type (EQL X). It'd
772 ;; be easy to transform one to the other, but it'd be somewhat
773 ;; messier to do so while ensuring that the EQL-SPECIALIZER-FORM
774 ;; is only EVAL'd once. (The new code wouldn't be messy, but
775 ;; it'd require a big transformation of the old code.) So
776 ;; instead we punt. -- WHN 20000610
780 ;; Parsed specializer objects, i.e. CLASS, EQL-SPECIALIZER,
781 ;; CLASS-EQ-SPECIALIZER and generic SPECIALIZER.
782 ;;
783 ;; Also unparsed specializers other than EQL: these have to be
784 ;; either class names or extended specializers.
785 ;;
786 ;; For these, we can usually make Python very happy.
787 ;;
788 ;; KLUDGE: Since INFO doesn't work right for class objects here,
789 ;; and they are valid specializers, see if the specializer is
790 ;; a named class, and use the name in that case -- otherwise
791 ;; the class instance is ok, since info will just return NIL, NIL.
792 ;;
793 ;; We still need to deal with the class case too, but at
794 ;; least #.(find-class 'integer) and integer as equivalent
795 ;; specializers with this.
798 proto-generic-function proto-method specializer))))))
805 ;; We have the %METHOD-NAME declaration in the place
806 ;; where we expect it only if there is are no
807 ;; non-standard prior MAKE-METHOD-LAMBDA methods -- or
808 ;; unless they're fantastically unintrusive.
811 ;; Macroexpansion caused by code-walking may call
812 ;; make-method-lambda and end up with wrong values
816 ;; the method-cell is a way of communicating what method
817 ;; a method-function implements, for the purpose of
818 ;; NO-NEXT-METHOD. We need something that can be shared
819 ;; between function and initargs, but not something that
820 ;; will be coalesced as a constant (because we are
821 ;; naughty, oh yes) with the expansion of any other
822 ;; methods in the same file. -- CSR, 2007-05-30
834 ;; These declarations seem to be used by PCL to pass
835 ;; information to itself; when I tried to delete 'em
836 ;; ca. 0.6.10 it didn't work. I'm not sure how they
837 ;; work, but note the (VAR-DECLARATION '%CLASS ..)
838 ;; expression in CAN-OPTIMIZE-ACCESS1. -- WHN
839 ;; 2000-12-30
843 parameters specializers)
844 ;; These TYPE declarations weren't in the original PCL
845 ;; code, but the Python compiler likes them a
846 ;; lot. (We're telling the compiler about our
847 ;; knowledge of specialized argument types so that it
848 ;; can avoid run-time type dispatch overhead, which
849 ;; can be a huge win for Python.)
853 proto-gf proto-method par spec specials env))
854 parameters specializers))))
859 required-parameters)))
861 ;; Remove the documentation string and insert the
862 ;; appropriate class declarations. The documentation
863 ;; string is removed to make it easy for us to insert
864 ;; new declarations later, they will just go after the
865 ;; CADR of the method lambda. The class declarations
866 ;; are inserted to communicate the class of the method's
867 ;; arguments to the code walk.
869 ;; The default ignorability of method parameters
870 ;; doesn't seem to be specified by ANSI. PCL had
871 ;; them basically ignorable but was a little
872 ;; inconsistent. E.g. even though the two
873 ;; method definitions
874 ;; (DEFMETHOD FOO ((X T) (Y T)) "Z")
875 ;; (DEFMETHOD FOO ((X T) Y) "Z")
876 ;; are otherwise equivalent, PCL treated Y as
877 ;; ignorable in the first definition but not in the
878 ;; second definition. We make all required
879 ;; parameters ignorable as a way of systematizing
880 ;; the old PCL behavior. -- WHN 2000-11-24
882 ,class-declarations
883 ,parameter-declarations
884 ,@declarations
903 method-lambda required-parameters env slots))
912 walked-lambda-body
914 ,slot-name-lists
924 aux))))
927 (,lambda-list .method-args. .next-methods.
928 :call-next-method-p
934 ,@walked-declarations
936 %parameter-binding-modified))
940 %parameter-binding-modified))
952 proto-generic-function proto-method name environment)))
963 :format-control
964 "~@<~S is not a valid parameter specializer name.~@:>"
973 specializer-name))
983 ;; In case of unknown specializer or known specializer with syntax
984 ;; error, TYPECASE may fall through to default method with error
985 ;; signaling.
1002 `(,type
1003 (,function-name
1004 proto-generic-function proto-method specializer-name environment))))
1006 ; ^ apparently not needed during bootstrapping
1007 symbol cons t)))))
1017 ;; TODO later protocol-unimplemented-error?
1020 nil)
1024 "~@<Cannot parse specializer ~S in ~S~@[: ~A~].~@:>"
1028 :format-control
1029 "~@<Cannot find type for specializer ~
1030 ~/sb-ext:print-symbol-with-prefix/ when executing ~S ~
1031 for a ~/sb-ext:print-type-specifier/ of a ~
1032 ~/sb-ext:print-type-specifier/.~@:>"
1033 :format-arguments
1041 (:primitive
1043 name
1045 name proto-generic-function proto-method)))
1046 (:defined
1047 ;; This can happen if NAME is a DEFTYPE.
1049 name proto-generic-function proto-method))
1051 ;; CLOS classes are too expensive to check (as opposed
1052 ;; to STRUCTURE-CLASS and SYSTEM-CLASS).
1053 nil)
1055 ;; TODO proper warning condition?
1057 name proto-generic-function proto-method)
1058 nil)))))
1060 ;;; Non-parsed class specializers, i.e. class names
1061 ;;;
1062 ;;; Extended generic function classes with specializers which are
1063 ;;; designated by symbols have to install their own methods
1064 ;;; specialized on symbol to replace this logic.
1070 ;; Usually tries to find the class named
1071 ;; SPECIALIZER. Can do something different when there
1072 ;; is a non-default method on
1073 ;; PARSE-SPECIALIZER-USING-CLASS.
1075 proto-generic-function specializer)
1077 ;; SPECIALIZER does not name a class, but maybe it is
1078 ;; known to name a :forthcoming-defclass-type.
1079 ;; CLASS-NAME-TYPE-SPECIFIER will emit the warning and
1080 ;; return nil if not.
1082 specializer proto-generic-function proto-method nil))
1084 ;; This can only happen if there is an EQL-specialized
1085 ;; method on PARSE-SPECIALIZER-USING-CLASS matching
1086 ;; SPECIALIZER that signals an error.
1088 nil))))
1091 proto-generic-function proto-method specializer))))
1093 ;;; Non-parsed extended specializer with default syntax
1094 ;;; i.e. (SPECIALIZER-KIND &rest SPECIFIC-SYNTAX)
1101 proto-generic-function specializer)
1103 ;; This can happen, for example, if SPECIALIZER does
1104 ;; not designate any extended specializer or if it does
1105 ;; but then does not conform to the respective extended
1106 ;; specializer syntax.
1108 nil))))
1111 proto-generic-function proto-method specializer))))
1113 ;;; Parsed EQL and CLASS-EQ specializers
1125 ;;; Parsed class specializers
1140 ;; Make sure SPECIALIZER has a proper class name and that name
1141 ;; designates the class SPECIALIZER in the global environment.
1145 name proto-generic-function proto-method)))))
1157 `(,type
1158 (,function-name
1159 proto-generic-function proto-method specializer))))
1161 structure-class system-class class)))))
1171 result
1173 specializer generic-function))))
1181 ;; FIXME: this HANDLER-CASE is a bit of a hammer to crack a nut:
1182 ;; the idea is that we want to unparse permissively, so that the
1183 ;; lazy (or rather the "portable") specializer extender (who
1184 ;; does not define methods on these new SBCL-specific MOP
1185 ;; functions) can still subclass specializer and define methods
1186 ;; without everything going wrong. Making it cleaner and
1187 ;; clearer that that is what we are defending against would be
1188 ;; nice. -- CSR, 2007-06-01
1195 class-name
1196 type))
1197 type))
1200 specializer generic-function)))
1206 ;;; For passing a list (groveled by the walker) of the required
1207 ;;; parameters whose bindings are modified in the method body to the
1208 ;;; optimized-slot-value* macros.
1212 method-args
1213 next-methods
1224 next-method-call
1225 args
1226 rest-arg
1235 parameters-setqd applyp method-cell))
1236 &body body
1240 ;; always provide the lexical function NEXT-METHOD-P.
1241 ;; I would think this to be a good candidate for declaring INLINE
1242 ;; but that's not the way it was done before.
1253 `((%check-cnm-args cnm-args
1254 ,method-args
1256 nil)
1264 ',method-cell
1273 ;; Can't easily provide a RETRY restart here, as the return value here is
1274 ;; for the method, not the generic function.
1276 method args)))
1294 call-method-args)
1296 value)
1317 pv
1318 next-method-call
1319 arg-info)
1322 value)
1326 ;; The two variants of INVOKE-FAST-METHOD-CALL differ in how REST-ARGs
1327 ;; are handled. The first one will get REST-ARG as a single list (as
1328 ;; the last argument), and will thus need to use APPLY. The second one
1329 ;; will get them as a &MORE argument, so we can pass the arguments
1330 ;; directly with MULTIPLE-VALUE-CALL and %MORE-ARG-VALUES.
1339 more-context
1340 more-count
1346 ,@required-args
1349 ;; The cases with only small amounts of required arguments passed
1350 ;; are probably very common, and special-casing speeds them up by
1351 ;; a factor of 2 with very little effect on the other
1352 ;; cases. Though it'd be nice to have the generic case be equally
1353 ;; fast.
1360 ,@required-args
1371 \f
1372 ;;;; effective method functions
1378 ;;; This function was in the CMU CL version of PCL (ca Debian 2.4.8)
1379 ;;; without explanation. It appears to be intended for debugging, so
1380 ;;; it might be useful someday, so I haven't deleted it.
1381 ;;; But it isn't documented and isn't used for anything now, so
1382 ;;; I've conditionalized it out of the base system. -- WHN 19991213
1383 #+sb-show
1416 ,@more-arg
1419 ,restp
1420 ,@required-args
1425 ;; "What," you may wonder, "do these next two clauses do?" In that
1426 ;; case, you are not a PCL implementor, for they considered this to
1427 ;; be self-documenting.:-| Or CSR, for that matter, since he can
1428 ;; also figure it out by looking at it without breaking stride. For
1429 ;; the rest of us, though: From what the code is doing with .SLOTS.
1430 ;; and whatnot, evidently it's implementing SLOT-VALUEish and
1431 ;; GET-SLOT-VALUEish things. Then we can reason backwards and
1432 ;; conclude that setting EMF to a FIXNUM is an optimized way to
1433 ;; represent these slot access operations.
1444 value)))))
1452 ;; (In cmucl-2.4.8 there was a commented-out third ,@(WHEN
1453 ;; ...) clause here to handle SLOT-BOUNDish stuff. Since
1454 ;; there was no explanation and presumably the code is 10+
1455 ;; years stale, I simply deleted it. -- WHN)
1456 )))
1458 ;;; Before SBCL 0.9.16.7 instead of
1459 ;;; INVOKE-NARROW-EFFECTIVE-METHOD-FUNCTION we passed a (THE (OR
1460 ;;; FUNCTION METHOD-CALL FAST-METHOD-CALL) EMF) form as the EMF. Now,
1461 ;;; to make less work for the compiler we take a path that doesn't
1462 ;;; involve the slot-accessor clause (where EMF is a FIXNUM) at all.
1477 ,@more-arg
1480 ,restp
1481 ,@required-args
1485 emf-n restp required-args))
1508 args)
1525 args))))))
1528 args
1538 value)))
1551 \f
1554 method-cell
1555 cnm-args)
1558 ,next-method-call
1570 ,@args
1576 call-next-method-p
1577 setq-p
1578 parameters-setqd
1579 method-cell
1580 applyp))
1581 &body body
1601 nil)
1603 ,next-method-call
1605 ,method-cell
1606 cnm-args))))
1612 ;;; CMUCL comment (Gerd Moellmann):
1613 ;;;
1614 ;;; The standard says it's an error if CALL-NEXT-METHOD is called with
1615 ;;; arguments, and the set of methods applicable to those arguments is
1616 ;;; different from the set of methods applicable to the original
1617 ;;; method arguments. (According to Barry Margolin, this rule was
1618 ;;; probably added to ensure that before and around methods are always
1619 ;;; run before primary methods.)
1620 ;;;
1621 ;;; This could be optimized for the case that the generic function
1622 ;;; doesn't have hairy methods, does have standard method combination,
1623 ;;; is a standard generic function, there are no methods defined on it
1624 ;;; for COMPUTE-APPLICABLE-METHODS and probably a lot more of such
1625 ;;; preconditions. That looks hairy and is probably not worth it,
1626 ;;; because this check will never be fast.
1628 ;; 1. Check for no arguments.
1633 ;; 2. Requirement arguments pairwise: if all are EQL, the applicable
1634 ;; methods must be the same. This takes care of the relatively common
1635 ;; case of twiddling with &KEY arguments without being horribly
1636 ;; expensive.
1643 ;; 3. Only then do the full check.
1648 ~{~S~^, ~} to call-next-method is different from ~
1649 the set of methods ~S applicable to the original ~
1650 method argument~P ~{~S~^, ~}.~@:>"
1654 ;; FIXME: replacing this entire mess with DESTRUCTURING-BIND would correct
1655 ;; problems similar to those already solved by a correct implementation
1656 ;; of DESTRUCTURING-BIND, such as incorrect binding order:
1657 ;; e.g. (macroexpand-1 '(bind-args ((&optional (x nil xsp)) args) (form)))
1658 ;; -> (LET* ((.ARGS-TAIL. ARGS) (XSP (NOT (NULL .ARGS-TAIL.))) (X ...)))
1659 ;; It's mostly irrelevant unless a method uses CALL-NEXT-METHOD though.
1675 "encountered the non-standard lambda list keyword ~S"
1676 var)))
1677 nil)
1725 ,@bindings
1726 (.dummy0.
1737 ;; FIXME: Do we want to export this symbol? Or maybe use an
1738 ;; (ERROR 'SIMPLE-PROGRAM-ERROR) form?
1741 return tail))
1745 ;; method definition
1747 ;; a list of all required parameters whose bindings might be
1748 ;; modified in the method body.
1755 ;; hierarchy: nil -> :simple -> T.
1758 form)
1760 ;; The walker will split (SETQ A 1 B 2) to
1761 ;; separate (SETQ A 1) and (SETQ B 2) forms, so we
1762 ;; only need to handle the simple case of SETQ
1763 ;; here.
1768 ;; Note that we don't need to check for
1769 ;; %VARIABLE-REBINDING declarations like is
1770 ;; done in CAN-OPTIMIZE-ACCESS1, since the
1771 ;; bindings that will have that declation will
1772 ;; never be SETQd.
1774 ;; If a parameter binding is shadowed by
1775 ;; another binding it won't have a
1776 ;; %PARAMETER declaration anymore, and this
1777 ;; won't get executed.
1779 form)
1783 form)
1791 form))
1795 ;;; FIXME: the walker's rewriting of the source code causes
1796 ;;; trouble when doing code coverage. The rewrites should be
1797 ;;; removed, and the same operations done using
1798 ;;; compiler-macros or tranforms.
1800 walked-lambda
1801 method-lambda)
1802 call-next-method-p
1804 parameters-setqd)))))
1812 \f
1822 new-value)
1824 \f
1830 #+nil
1834 ll initargs source-location)))
1838 initargs source-location)
1847 :name gf-spec
1848 :new-location source-location
1849 :old-method method
1852 gf-spec qualifiers specializers lambda-list
1853 'source source-location
1854 initargs)))
1857 ;; FIXME: should be STYLE-WARNING?
1859 "~&At the time the method with qualifiers ~:S and~%~
1860 specializers ~:S on the generic function ~S~%~
1861 was compiled, the method-class for that generic function was~%~
1862 ~S. But, the method class is now ~S, this~%~
1863 may mean that this method was compiled improperly.~%"
1864 qualifiers specializers gf-spec
1866 method))
1894 \f
1897 ;; We say "&MUMBLE is not allowed in a generic function lambda list"
1898 ;; whether this is called by DEFMETHOD or DEFGENERIC.
1899 ;; [It is used for either. Why else recognize and silently ignore &AUX?]
1903 :silent t
1909 ;; FIXME: this does more than return an FTYPE from a lambda list -
1910 ;; it unions the type with an existing ctype object. It needs a better name,
1911 ;; and to be reimplemented as "union and call sb-c::ftype-from-lambda-list".
1922 old-ftype))))
1937 keywords)
1940 *))))
1941 \f
1942 ;;;; early generic function support
1946 ;; CLHS doesn't specify &allow-other-keys here but I guess the supposition
1947 ;; is that they'll be checked by ENSURE-GENERIC-FUNCTION-USING-CLASS.
1948 ;; Except we don't do that either, so I think the blame, if any, lies there
1949 ;; for not catching errant keywords.
1961 existing fun-name all-keys)))))
1965 'simple-program-error
1967 already names an ordinary function or a ~
1968 macro.~@:>"
2017 arg-info-precedence
2018 arg-info-metatypes
2019 arg-info-number-optional
2020 arg-info-key/rest-p
2021 arg-info-keys ;nil no &KEY or &REST allowed
2022 ;(k1 k2 ..) Each method must accept these &KEY arguments.
2023 ;T must have &KEY or &REST
2025 gf-info-simple-accessor-type ; nil, reader, writer, boundp
2028 gf-info-static-c-a-m-emf
2030 gf-info-fast-mf-p)
2048 ;;; Create a gf lambda list from a method lambda list
2061 argument-precedence-order)
2086 existing methods of ~S.~%~
2087 Old lambda-list ~s"
2094 lambda-list
2110 arg-info))
2119 to the generic function~2I~_~S;~I~_ but ~
2120 ~?~:>"
2121 method gf string args))
2130 "the method has ~A required arguments than the generic function."
2134 "the method has ~A optional arguments than the generic function."
2138 "the method and generic function differ in whether they accept~_~
2145 ~S."
2146 gf-keywords)))))))
2155 ;;; FIXME: we don't actually need this; we could test for the exact
2156 ;;; class and deal with it as appropriate. In fact we probably don't
2157 ;;; need it anyway because we only use this for METHOD-SPECIALIZERS on
2158 ;;; the standard reader method for METHOD-SPECIALIZERS. Probably.
2170 standard-writer-method standard-boundp-method
2171 global-reader-method global-writer-method
2172 global-boundp-method)
2203 nil)))
2251 nil)
2257 ;; FIXME: this test will eventually be
2258 ;; superseded by the *internal-pcl...* test,
2259 ;; above. While we are in a process of
2260 ;; transition, however, it should probably
2261 ;; remain.
2267 #'make-method-lambda
2278 arg-info)
2280 ;;; This is the early definition of ENSURE-GENERIC-FUNCTION-USING-CLASS.
2281 ;;;
2282 ;;; The STATIC-SLOTS field of the funcallable instances used as early
2283 ;;; generic functions is used to store the early methods and early
2284 ;;; discriminator code for the early generic function. The static
2285 ;;; slots field of the fins contains a list whose:
2286 ;;; CAR - a list of the early methods on this early gf
2287 ;;; CADR - the early discriminator code for this method
2290 lambda-list-p)
2291 argument-precedence-order
2293 documentation
2299 existing)
2303 argument-precedence-order source
2304 documentation)
2312 argument-precedence-order source
2313 documentation))))
2316 function argument-precedence-order source-location
2317 documentation)
2337 :lambda-list lambda-list
2340 fin))
2350 new-value)
2356 dfun)))
2359 ;; Check that we are under the lock.
2360 #+sb-thread
2365 new-state))))
2366 dfun)
2401 nil)
2411 ;; Use the existing lambda-list, if any. It is reasonable to do eg.
2412 ;;
2413 ;; (if (fboundp name)
2414 ;; (ensure-generic-function name)
2415 ;; (ensure-generic-function name :lambda-list '(foo)))
2416 ;;
2417 ;; in which case we end up here with no lambda-list in the first leg.
2421 t)
2430 ;; FIXME: Ideally we would like to not clobber it, but because generic
2431 ;; functions assert their FTYPEs callers believing the FTYPE are left with
2432 ;; unsafe assumptions. Hence the clobbering. Be quiet when the new type
2433 ;; is a subtype of the old one, though -- even though the type is not
2434 ;; trusted anymore, the warning is still not quite as interesting.
2438 ~/sb-ext:print-symbol-with-prefix/ clobbers an ~
2439 earlier ~S proclamation ~/sb-ext:print-type/ for ~
2440 the same name with ~/sb-ext:print-type/.~:@>"
2444 fun-name))
2450 class-or-name)
2453 symbol that names a class.~@:>"
2454 context class-or-name))))
2457 class-or-name environment)))
2459 ;; FIXME: reference MOP documentation -- this is an
2460 ;; additional requirement on our users
2462 finalizeable~@:>"
2463 class)
2466 environment
2471 generic-function-class environment)))
2476 (:method-combination
2478 key
2485 value))))
2486 (:method-class
2488 value environment)))
2515 \f
2522 ;;; FIXME: this function took on a slightly greater role than it
2523 ;;; previously had around 2005-11-02, when CSR fixed the bug whereby
2524 ;;; having more than one subclass of standard-generic-function caused
2525 ;;; the whole system to die horribly through a metacircle in
2526 ;;; GF-ARG-INFO. The fix is to be slightly more disciplined about
2527 ;;; calling accessor methods -- we call GET-GENERIC-FUN-INFO when
2528 ;;; computing discriminating functions, so we need to be careful about
2529 ;;; having a base case for the recursion, and we provide that with the
2530 ;;; STANDARD-GENERIC-FUNCTION case below. However, we are not (yet)
2531 ;;; as disciplined as CLISP's CLOS/MOP, and it would be nice to get to
2532 ;;; that stage, where all potentially dangerous cases are enumerated
2533 ;;; and stopped. -- CSR, 2005-11-02.
2535 ;; values nreq applyp metatypes nkeys arg-info
2542 metatypes
2543 arg-info))
2552 nkeys
2553 arg-info))))
2564 &key slot-name object-class method-class-function
2568 ;; Figure out whether we got class objects or class names as the
2569 ;; specializers and set parsed and unparsed appropriately. If we
2570 ;; got class objects, then we can compute unparsed, but if we
2571 ;; got class names we don't try to compute parsed.
2577 specializers))
2588 ;; the parsed specializers. This is used by
2589 ;; EARLY-METHOD-SPECIALIZERS to cache the parse.
2590 ;; Note that this only comes into play when there is
2591 ;; more than one early method on an early gf.
2592 parsed
2594 ;; A list to which REAL-MAKE-A-METHOD can be applied
2595 ;; to make a real method corresponding to this early
2596 ;; one.
2599 initargs doc)
2605 result))
2621 :slot-definition slot-definition
2625 initargs)
2626 'source source
2627 initargs)))
2647 ;;; Fetch the specializers of an early method. This is basically just
2648 ;;; a simple accessor except that when the second argument is t, this
2649 ;;; converts the specializers from symbols into class objects. The
2650 ;;; class objects are cached in the early method, this makes
2651 ;;; bootstrapping faster because the class objects only have to be
2652 ;;; computed once.
2653 ;;;
2654 ;;; NOTE:
2655 ;;; The second argument should only be passed as T by
2656 ;;; early-lookup-method. This is to implement the rule that only when
2657 ;;; there is more than one early method on a generic function is the
2658 ;;; conversion from class names to class objects done. This
2659 ;;; corresponds to the fact that we are only allowed to have one
2660 ;;; method on any generic function up until the time classes exist.
2685 specializers arglist &rest initargs
2689 ;; argument here because the default,
2690 ;; STANDARD-GENERIC-FUNCTION, is right for all early generic
2691 ;; functions. (See REAL-ADD-NAMED-METHOD)
2701 specializers initargs documentation
2706 ;;; This is the early version of ADD-METHOD. Later this will become a
2707 ;;; generic function. See !FIX-EARLY-GENERIC-FUNCTIONS which has
2708 ;;; special knowledge about ADD-METHOD.
2721 ;;; This is the early version of REMOVE-METHOD. See comments on
2722 ;;; the early version of ADD-METHOD.
2736 ;;; This is the early version of GET-METHOD. See comments on the early
2737 ;;; version of ADD-METHOD.
2743 specializers)
2745 specializers))
2750 nil))
2753 ;; minor KLUDGE: a separate code component for this function allows GCing
2754 ;; a few symbols and their associated code that would otherwise be retained:
2755 ;; *!EARLY-{GENERIC-}FUNCTIONS*, *!GENERIC-FUNCTION-FIXUPS*
2763 ;; Rearrange *!EARLY-GENERIC-FUNCTIONS* to speed up
2764 ;; FIX-EARLY-GENERIC-FUNCTIONS.
2771 generic-function-methods
2772 method-specializers
2773 specializer-type
2774 specializer-class
2775 slot-definition-location
2776 slot-definition-name
2777 class-slots
2778 gf-arg-info
2779 class-precedence-list
2780 slot-boundp-using-class
2782 slot-value-using-class)))
2793 early-method))))
2796 early-method t))
2811 ;; This is lifted from sb-introspect, OAOO and all that.
2830 map)
2844 'standard-method
2845 qualifiers lambda-list specializers initargs nil
2856 \f
2857 ;;; PARSE-DEFMETHOD is used by DEFMETHOD to parse the &REST argument
2858 ;;; into the 'real' arguments. This is where the syntax of DEFMETHOD
2859 ;;; is really implemented.
2881 \f
2885 ;; We don't need these, but according to the unit tests,
2886 ;; they're mandated by AMOP.
2892 ()
2897 :format-control format-control
2900 ;; Return 3 values:
2901 ;; - the bound variables, without defaults, supplied-p vars, or &AUX vars.
2902 ;; - the lambda list without specializers.
2903 ;; - just the specializers
2907 arglist
2916 llks required optional rest key aux nil nil
2920 rest
2921 ;; Preserve keyword-names when given as (:KEYWORD var)
2926 key))
2928 specializers)))
2929 \f
2931 \f
2932 ;;; FIXME: In here there was a #-CMU definition of SYMBOL-MACROLET
2933 ;;; which used %WALKER stuff. That suggests to me that maybe the code
2934 ;;; walker stuff was only used for implementing stuff like that; maybe
2935 ;;; it's not needed any more? Hunt down what it was used for and see.
2941 form context))
2944 form)))
2957 ,in
2965 either a symbol or (variable-name ~
2966 slot-name)"
2967 slot-entry))
2971 `(,var-name
2973 slots)
2981 ,in
2987 be (variable-name accessor-name)"
2988 slot-entry))
2990 slot-entry
2992 slots)