friedman@porthos.rutgers.edu (Gadi ) (11/03/88)
I am using
Allegro CL 3.0.1.beta [sun4] (7/29/88 23:05)
Copyright (C) 1985-1988, Franz Inc., Berkeley, CA, USA
and am tring to setf array values using (apply #'aref array list))
(setq test (make-array '(2 2))) ; works
(apply #'aref test '(1 1)) ; works (returns the correct element)
(setf (apply #'aref test '(1 1)) 2) ; DOES NOT WORK.
This should work, as described in CLtL pg 95 and 291.
It also works fine using
;;; Sun Common Lisp, Development Environment 2.1.1, 24-Jun-88
;;;
;;; Copyright (c) 1987 by Sun Microsystems, Inc. All Rights Reserved
;;; Copyright (c) 1985, 1986, 1987 by Lucid Inc., All Rights Reserved
Can anyone think of a workaround. Something short of making a
new list and using eval. (eval (setf (aref test 1 1) 2))
will work. (We all know how slow eval is)
--
uucp: {ames, cbosgd, harvard, moss}!rutgers!aramis.rutgers.edu!friedman
arpa: FRIEDMAN@ARAMIS.RUTGERS.EDUhedrick@geneva.rutgers.edu (Charles Hedrick) (11/03/88)
Unless you really don't know the number of dimensions of your array in advance, (apply #'aref array list) sounds to me like it's going to be inefficient. It's going to be particularly hard to produce good compiled code for it. I agree that apply is better than eval, but not by that much. At any rate, the inverse of aref in Allegro CL appears to be excl::.inv-s-aref. So you can probably do (apply #'excl::.inv-s-aref newvalue array subscript-list) This will work until we get Franz to make (setf (apply #'aref ... work.
layer@snooze.UUCP (Kevin Layer) (11/08/88)
++ From: friedman@porthos.rutgers.edu (Gadi ) ++ ++ I am using ++ ++ Allegro CL 3.0.1.beta [sun4] (7/29/88 23:05) ++ Copyright (C) 1985-1988, Franz Inc., Berkeley, CA, USA ++ ++ and am tring to setf array values using (apply #'aref array list)) ++ ++ (setq test (make-array '(2 2))) ; works ++ (apply #'aref test '(1 1)) ; works (returns the correct element) ++ (setf (apply #'aref test '(1 1)) 2) ; DOES NOT WORK. ++ This should work, as described in CLtL pg 95 and 291. ++ There is patch for this bug, which will be sent to you under separate cover. ++ From: hedrick@geneva.rutgers.edu (Charles Hedrick) ++ ++ Unless you really don't know the number of dimensions of your array in ++ advance, (apply #'aref array list) sounds to me like it's going to be ++ inefficient. It's going to be particularly hard to produce good ++ compiled code for it. I agree that apply is better than eval, but not ++ by that much. At any rate, the inverse of aref in Allegro CL appears ++ to be excl::.inv-s-aref. So you can probably do ++ ++ (apply #'excl::.inv-s-aref newvalue array subscript-list) ++ ++ This will work until we get Franz to make (setf (apply #'aref ... ++ work. You are correct, this is the proper workaround in the absence of the patch. Kevin Layer Franz Inc.