|
From: | Richard Fateman |
Subject: | [Gcl-devel] solve test on allegro CL maxima |
Date: | Thu, 21 Mar 2002 10:30:36 -0800 |
User-agent: | Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:0.9.4) Gecko/20011128 Netscape6/6.2.1 |
(C23) W:SOLVE([4*X^2-Y^2 = 12,X*Y-X = 2],[X,Y]) ; Error: STORE failed -- can't find array for #<macro AFIXN @ #x203d5f12> The backtrace in Allegro says...(ERROR "STORE failed -- can't find array for ~S" #<macro AFIXN @ #x203d5f12>)
->(STORE-INTERNAL-2D #<macro AFIXN @ #x203d5f12> 1 ...) (CPTOMF 5 (#:X1374 3 1 ...) ...) (CPBQ1 (#:X1374 3 1 ...) 3) (CPBER1 (#:X1374 4 1 ...)) (CHOOZP (#:X1374 4 4 ...)) (FACT5 (#:X1374 4 4 ...)) (FACTOR72 (#:X1374 4 4 ...)) (FACTOR1972 (#:X1374 4 4 ...)) So it is actually in the polynomial factoring program. I think this may be trying factoring over Gaussian integers. This may or may not be related to GCL problems. For what it is worth, the commercial macsyma gives the following (better, in my view) answer:[[x = 2, y = 2], [x = root_of(4 * x^3 + 8 * x^2 + 3 * x + 2), y = ((x + 2)/x)]]
RJF Camm Maguire wrote:
Greetings! "Vadim V. Zhytnikov" <address@hidden> writes:Some more facts about the problem I was able to get lately: 1) The problem is not due to recent solve changes. It exists with older solve as well. I'm not surprised but I verified it explicitly. 2) The problem is not due to recent GCL updates. It shows up both with GCL 2.5.0 (latest CVS) and old GCL 2.3.8. 3) When I start tracing some functions (using lisp's trace) in which error occurs the problem magically disappears! This is very bad and makes detection of the problem's origin rather hard. Something really nasty is happening to variables binding.Yes, please see my other post about similar mysterious behavior which I feel arises from the same malady. to this list, I might also add again the note that certain catastrophic problems on m68k Linux were due to missing casts in the function call pointer routines. This is making me conclude that we are ready for a straightforward but tedious task for reworking all of the C files to compile with no warnings under -Wall. A useful tool for this is protoize, for those who might not be aware. at the same time, we ought to modernize the varargs paramters specifications, and use ansi C function declarations. We probably will also have to make minor modifications to the lisp compiler to get the generated C code it produces into thesame form.Any suggestions as to how to organize this task is appreciated. I feel very comfortable with plodding along in this direction, but as by now is obvious, my time is limited, and it will take a considerable amount of time. I hope to be able to clear a day for gcl by sometime next week. Take care,Vadim Camm Maguire wrote:Greetings! "Vadim V. Zhytnikov" <address@hidden> writes:This is a multi-part message in MIME format. --------------DCA8AA4BB1B26CA471011469 Content-Type: text/plain; charset=koi8-r Content-Transfer-Encoding: 7bit Greetings! Finally I located the problem with solve test on GCL Maxima build with new build system.Does this error disappear with the old build system?If you load("test.lisp"); the following file -------- tests.lisp --------- (sloop for v in '("rtest.mac") do (test-batch v) ) ----------------------------- with ------- rtest.mac ---------- W:SOLVE([4*X^2-Y^2 = 12,X*Y-X = 2],[X,Y]); [[X = 2,Y = 2], [X = 0.5202594388652008*%I-0.1331240357358706, Y = 0.07678378523787777-3.608003221870287*%I], [X = -0.5202594388652008*%I-0.1331240357358706, Y = 3.608003221870287*%I+0.07678378523787777], [X = -1.733751846381093,Y = -0.1535675710019696]]; ----------------------------- in GCL Maxima you will get an error. But it is enough to change sloop variable v to something different to get correct answer! On Clisp everything works fine independently on sloop variable name.Well, unfortunately, I suspect gcl here. Which version? What about different input functions? Can you go through the lisp debugger and isolate the lisp error? I'd really appreciate it if you could get this into a form where a simple (as possible) input file into gcl shows the mistake. Otherwise, we can of course build the gcl with -g and step through the stack in gdb, but I'm sure its quite huge.I'm perfectly sure that the problem has nothing to do with recent solve changes. It is not clear yet which part is to be blamed for the problem - new build system or GCL. The case require more thorough testing.Thanks again for your work here, Vadim!Vadim -- [ Vadim V. Zhytnikov <address@hidden> <address@hidden> ] --------------DCA8AA4BB1B26CA471011469 Content-Type: application/x-gzip; name="solve.tar.gz" Content-Transfer-Encoding: base64 Content-Disposition: inline; filename="solve.tar.gz" H4sIAM/8eTwAA+3UXWvCMBQG4N6uvyIIY62z5SQnX052uYvBYBeDrUUUnG5M0Cna+fsXKzgG buKF+4D3uUjT9JDTkJxUT8sqn4yX8+h4SBJZrSMKnDWfnmtSUUROs2WnNXOIl45lJOiI/7T1 tqwGCyGi1WA0nn4Tt+/7P5UsJ7PZXDzPFmIlxq/iLGksqvWZmA6GjTQ+EaNZaJL1UPY4qIYv YpXGafyl314PHGa72UfMsa/+tVWb+leGnLR1/UuF+v8JDxd3tzf3V0lXN4u+ysq+EpdCqlbR LLMidFWv1S1aZS/txN1uPdAqN8OxqN8pN4qUaWv23oYu+ebpdUa5ZJZKExvHxjuyIV6UdXzY e+fZeaNC44KMc0ueiJWSIVR5F6bYJsgOzbBrtvMdaT8yyNwxOyO9tuwltble4zqFYWOdceEA y7Zt216vgxsOAAAAAAAAAAAAAAAAAAAA/ph3Wq1t4QAoAAA= --------------DCA8AA4BB1B26CA471011469---- Camm Maguire address@hidden ========================================================================== "The earth is but one country, and mankind its citizens." -- Baha'u'llah _______________________________________________ Gcl-devel mailing list address@hidden http://mail.gnu.org/mailman/listinfo/gcl-devel-- [ [ Vadim V. Zhytnikov [ [ <address@hidden> [ <address@hidden> [ <address@hidden> [
[Prev in Thread] | Current Thread | [Next in Thread] |