[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnucap-devel] Re: revsion control
From: |
al davis |
Subject: |
Re: [Gnucap-devel] Re: revsion control |
Date: |
Tue, 12 Sep 2006 19:01:34 -0400 |
User-agent: |
KMail/1.9.4 |
On Friday 08 September 2006 03:16, Jonathan David wrote:
> I'll also suggest that we setup a trac wiki/bug tracker for
> the project. Are you coming out a couple day's before BMAS?
> Maybe I have someone who can help set up that infrastructure.
> Jonathan
I have a wiki set up at gnucap.org, but there is nothing on it
yet. It is running dokuwiki, chosen mainly because that is the
one that gEDA uses, which should make it easy to mirror there.
Since there is nothing there yet, the choice is not firm.
Formal bug tracking is probably a good idea, but it has not been
a problem. The number is bugs is low compared to most
projects, and most get fixed immediately when they are
discovered. There are built-in checks in the code for many
potential problems. The code not checked adequately by the
regression suite is documented. Almost all bug reports
received so far are related to a user error. In this case, the
bug (or bugs) are usually in the error handling code, which is
tested less rigorously than the mainline code. Gnucap uses
structured exception handling, which makes error handling much
easier.
Another issue in a simulator is that the algorithms are not
perfect, sometimes causing trouble for the user. By this, I
mean there are inherent deficiencies that are not bugs, but
rather just the way the algorithm works. For example, it is
known that the Newton-Raphson method does not always converge.
Another example, it is known that certain time step issues can
show artifacts of numerical differentiation.
There is something set up at gnu.org. I have not used it yet.