From: "Yi Cao" <y.cao@cranfield.ac.uk>
Path: news.mathworks.com!newsfeed.mathworks.com!WebX
Newsgroups: comp.soft-sys.matlab
Subject: Re: All in good fun
Message-ID: <eeb4a65.90@WebX.raydaftYaTP>
Date: Sat, 16 Nov 2002 15:35:51 -0500
References: <eeb4a65.80@WebX.raydaftYaTP> <eeb4a65.82@WebX.raydaftYaTP> <eeb4a65.83@WebX.raydaftYaTP> <eeb4a65.84@WebX.raydaftYaTP> <eeb4a65.87@WebX.raydaftYaTP>
Lines: 14
MIME-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
Xref: news.mathworks.com comp.soft-sys.matlab:143158


Obviously, global variables have been cleared after each call to the
solver since the global hole was spotted. I noticed it because I
could not run some top entry code in my own computer. I found this is
because the code has an error using a global variable in the way,
C(:,i)=c without redefining correct size for C. Since these code can
pass master test, the master must clear global variables between each
call to solver. Someone suggested that the master test may have
sorted in size so that C can gradually expand without error. However,
after we see the master testsuit, this is not the case.


Bert Jagers wrote:
> work. But then one should also clear globals between successive
calls
> of the solver, otherwise one would be