| blob | 18d4727dd1b6f8ef10011643556e01c2d60155fc |
1 ;;;; cross-compiler-only versions of TYPEP, TYPE-OF, and related functions
3 ;;;; This software is part of the SBCL system. See the README file for
4 ;;;; more information.
5 ;;;;
6 ;;;; This software is derived from the CMU CL system, which was
7 ;;;; written at Carnegie Mellon University and released into the
8 ;;;; public domain. The software is in the public domain and is
9 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
10 ;;;; files for more information.
14 ;;; Is X a fixnum in the target Lisp?
19 ;;; (This was a useful warning when trying to get bootstrapping
20 ;;; to work, but it's mostly irrelevant noise now that the system
21 ;;; works.)
27 ))
30 s
31 "cross-compilation-time type ambiguity (should be OK) in ~S:~%~A"
35 ;;; This warning is issued when giving up on a type calculation where a
36 ;;; conservative answer is acceptable. Since a conservative answer is
37 ;;; acceptable, the only downside is lost optimization opportunities.
42 )))
44 ;;; This warning refers to the flexibility in the ANSI spec with
45 ;;; regard to run-time distinctions between floating point types.
46 ;;; (E.g. the cross-compilation host might not even distinguish
47 ;;; between SINGLE-FLOAT and DOUBLE-FLOAT, so a DOUBLE-FLOAT number
48 ;;; would test positive as SINGLE-FLOAT.) If the target SBCL does make
49 ;;; this distinction, then information is lost. It's not too hard to
50 ;;; contrive situations where this would be a problem. In practice we
51 ;;; don't tend to run into them because all widely used Common Lisp
52 ;;; environments do recognize the distinction between SINGLE-FLOAT and
53 ;;; DOUBLE-FLOAT, and we don't really need the other distinctions
54 ;;; (e.g. between SHORT-FLOAT and SINGLE-FLOAT), so we call
55 ;;; WARN-POSSIBLE-CROSS-TYPE-FLOAT-INFO-LOSS to test at runtime
56 ;;; whether we need to worry about this at all, and not warn unless we
57 ;;; do. If we *do* have to worry about this at runtime, my (WHN
58 ;;; 19990808) guess is that the system will break in multiple places,
59 ;;; so this is a real WARNING, not just a STYLE-WARNING.
60 ;;;
61 ;;; KLUDGE: If we ever try to support LONG-FLOAT or SHORT-FLOAT, this
62 ;;; situation will get a lot more complicated.
74 raw-result)
79 ;; it's a FIXNUM, but instead test against target
80 ;; MOST-fooITIVE-FIXNUM limits.
90 raw-result)
94 ;;; Is SYMBOL in the CL package? Note that we're testing this on the
95 ;;; cross-compilation host, which could do things any old way. In
96 ;;; particular, it might be in the CL package even though
97 ;;; SYMBOL-PACKAGE is not (FIND-PACKAGE :CL). So we test things
98 ;;; another way.
101 symbol))
103 ;; Return T if SYMBOL is a predicate acceptable for use in a SATISFIES type
104 ;; specifier. We assume that anything in CL: is allowed (see explanation at
105 ;; call point), and beyond that, anything we define has to be expressly listed
106 ;; here, for fear of later unexpected confusion.
110 ;; KLUDGE: rather than extensible list of predicates that match
111 ;; in behavior between the host and target lisp, hardcode a few.
115 ;;; This is like TYPEP, except that it asks whether HOST-OBJECT would
116 ;;; be of TARGET-TYPE when instantiated on the target SBCL. Since this
117 ;;; is hard to determine in some cases, and since in other cases we
118 ;;; just haven't bothered to try, it needs to return two values, just
119 ;;; like SUBTYPEP: the first value for its conservative opinion (never
120 ;;; T unless it's certain) and the second value to tell whether it's
121 ;;; certain.
125 ;; We don't have to keep track of this as long as system
126 ;; performance is acceptable, since giving up
127 ;; conservatively is a safe way out.
128 #+nil
135 ;; a convenient idiom for making more matches to special cases:
136 ;; Test both forms of target type for membership in LIST.
137 ;;
138 ;; (In order to avoid having to use too much deep knowledge
139 ;; of types, it's sometimes convenient to test RAW-TARGET-TYPE
140 ;; as well as the expanded type, since we can get matches with
141 ;; just EQL. E.g. SIMPLE-STRING can be matched with EQL, while
142 ;; safely matching its expansion,
143 ;; (OR (SIMPLE-ARRAY CHARACTER (*)) (SIMPLE-BASE-STRING *))
144 ;; would require logic clever enough to know that, e.g., OR is
145 ;; commutative.)
150 ;; the ANSI cross-compilation host. KLUDGE: This code will
151 ;; need to be tweaked by hand if the names of these types
152 ;; ever change, ugh!
161 ;; instead a CLASS object.
178 ;; partition, but it includes at least lists and vectors.
198 ;; KLUDGE: SBCL has * as an explicit wild type. While
199 ;; this is sort of logical (because (e.g. (ARRAY * 1)) is
200 ;; a valid type) it's not ANSI: looking at the ANSI
201 ;; definitions of complex types like like ARRAY shows
202 ;; that they consider * different from other type names.
203 ;; Someday we should probably get rid of this non-ANSIism
204 ;; in base SBCL, but until we do, we might as well here
205 ;; in the cross compiler. And in order to make sure that
206 ;; we don't continue doing it after we someday patch
207 ;; SBCL's type system so that * is no longer a type, we
208 ;; make this assertion. -- WHN 2001-08-08
212 ;; between any host ANSI Common Lisp and the target
213 ;; Common Lisp. (Some array types are too, but they
214 ;; were picked off earlier.)
217 list nil null number rational real signed-byte symbol t
218 unsigned-byte))
221 ;; too, but less exactly.
230 ;; Our dumping logic is based on contents, however, so
231 ;; reasoning about them should be safe
245 ;; We won't know how to dump it either.
249 ;; Some types require translation between the cross-compilation
250 ;; host Common Lisp and the target SBCL.
257 t))
258 ;; Some types are too hard to handle in the positive
259 ;; case, but at least we can be confident in a large
260 ;; fraction of the negative cases..
274 ;; Neither target CL:STREAM nor target SB!KERNEL:INSTANCE
275 ;; is implemented as a STRUCTURE-OBJECT, so they'll fall
276 ;; through the tests above. We don't want to assume too
277 ;; much about them here, but at least we know enough
278 ;; about them to say that neither T nor NIL nor indeed
279 ;; any other symbol in the cross-compilation host is one.
280 ;; That knowledge suffices to answer so many of the
281 ;; questions that the cross-compiler asks that it's well
282 ;; worth special-casing it here.
286 ;; various hacks for composite types..
291 ;; Many complex types are guaranteed to correspond exactly
292 ;; between any host ANSI Common Lisp and the target SBCL.
295 ;; Floating point types are guaranteed to correspond,
296 ;; too, but less exactly.
303 ;; Some complex types have translations that are less
304 ;; trivial.
313 ;; If we want to work with the KEYWORD type, we need
314 ;; to grok (SATISFIES KEYWORDP).
318 ;; Many predicates like KEYWORDP, ODDP, PACKAGEP,
319 ;; and NULL correspond between host and target.
320 ;; But we still need to handle errors, because
321 ;; the code which calls us may not understand
322 ;; that a type is unreachable. (E.g. when compiling
323 ;; (AND STRING (SATISFIES ARRAY-HAS-FILL-POINTER-P))
324 ;; CTYPEP may be called on the SATISFIES expression
325 ;; even for non-STRINGs.)
328 host-object))
332 ;; For symbols not in the CL package, it's not
333 ;; in general clear how things correspond
334 ;; between host and target, so we punt.
336 ;; Some complex types are too hard to handle in the
337 ;; positive case, but at least we can be confident in
338 ;; a large fraction of the negative cases..
355 ;; And the Common Lisp type system is complicated,
356 ;; and we don't try to implement everything.
358 ;; And the Common Lisp type system is complicated, and
359 ;; we don't try to implement everything.
363 ;;; This is an incomplete TYPEP which runs at cross-compile time to
364 ;;; tell whether OBJECT is the host Lisp representation of a target
365 ;;; SBCL type specified by TARGET-TYPE-SPEC. It need make no pretense
366 ;;; to completeness, since it need only handle the cases which arise
367 ;;; when building SBCL itself, e.g. testing that range limits FOO and
368 ;;; BAR in (INTEGER FOO BAR) are INTEGERs.
374 ;; A program that calls TYPEP doesn't want uncertainty and
375 ;; probably can't handle it.
377 opinion
379 host-object
380 target-type-spec))))
382 ;;; This is an incomplete, portable implementation for use at
383 ;;; cross-compile time only.
386 ;; There is at least one possible endless recursion in the
387 ;; cross-compiler type system: (SUBTYPEP NULL (OR UNKOWN0 UNKNOWN1)
388 ;; runs out of stack. The right way would probably be to not
389 ;; implement CTYPEP in terms of TYPE-SPECIFIER (:UNPARSE, that may
390 ;; call TYPE=, that in turn may call CTYPEP). Until then, pick a few
391 ;; cherries off.
399 ;; This is essentially just the ARRAY-TYPE case of %%TYPEP
400 ;; using !SPECIALIZED-ARRAY-ELEMENT-TYPE, not ARRAY-ELEMENT-TYPE.
428 ;; Since at cross-compile time we build a CLOS-free bootstrap
429 ;; version of SBCL, it's unclear how to explain to it what a
430 ;; generic function is.
432 ;; There's no ANSI way to find out what the function is
433 ;; declared to be, so we just return the CTYPE for the
434 ;; most-general function.
441 ;; It is critical not to inquire of the host for the array's element type.
444 ;; complexp relies on the host implementation,
445 ;; but in practice any array for which we need to
446 ;; call ctype-of will be a simple-array.
448 :element-type etype
453 ;; (Note that SBCL doesn't distinguish between BASE-CHAR and
454 ;; CHARACTER.)
457 nil)
459 ;; Beyond this, there seems to be no portable correspondence.
464 ;; There might be more cases which we could handle with
465 ;; sufficient effort; since all we *need* to handle are enough
466 ;; cases for bootstrapping, we don't try to be complete here,. If
467 ;; future maintainers make the bootstrap code more complicated,
468 ;; they can also add new cases here to handle it. -- WHN 2000-11-11