I've tried 5.9.0.rc4 on Linux with three lisps
(gcl, clisp and cmucl) and Windows binaries
on several machines. Unfortunately I observe
one problem with GCL maxima which was already
reported by Nokolaos I. Ioakimidis and amounts to
desolve test failure when tests are executed
in xmaxima. The problem is not so serious but
IMHO it must be fixed somehow before final
release since program which fails to pass its
own tests makes bad impression.
Details:
========================================================
/* ********************** Problem 41. *************** */
%Input is
EQN1 : f(x) = G(x) + SIN(x)
x x
The result is
(x + 2) = G(x) + SIN(x)
x x
..Which was correct
/* ********************** Problem 42. *************** */
%Input is
EQN2 : G(x) = f(x) - COS(x)
x x x
The result is
G(x) = (x + 2) - COS(x)
x x x
..Which was correct
/* ********************** Problem 43. *************** */
%Input is
DESOLVE([EQN1, EQN2], [f(x), G(x)])
x + 2 contains more than one independent variable.
-- an error. Quitting. To debug this try DEBUGMODE(TRUE);)
(C1) x;
(D1) x
(C2) eqn1;
(D2) (x + 2) = G(x) + SIN(x)
x x
(C3) eqn2;
(D3) G(x) = (x + 2) - COS(x)
x x x
(C4) f(x);
(D4) x + 2
(C5)
========================================================
It is obvious that the problem has nothing
to do with DESOLVE but induced by compiling
f(x):=x+2 function in rtest15.mac which
remains unkilled.
Quite strange but error shows up only on
some computers and OS'es - on others test
pass without any trouble (and it is OK
in text mode). My observations are:
Windows Xp, Me, 98SE on Intel processors
(Celeron, Pentium III) - FAILURE
Windows Xp, Me on Athlon XP - OK
Linux on Athlon XP - FAILURE
Strange pattern. But maybe we just
have to rename functions in DESOLVE
tests say f,g -> ff,gg. Of course
this just hides the trouble but
I don't want any delay 5.9.0 release.
--
Vadim V. Zhytnikov
<vvzhy@mail.ru>
<vvzhy@netorn.ru>