Subject: Re: CVS commit: basesrc
To: M Graff <explorer@flame.org>
From: Luke Mewburn <lukem@goanna.cs.rmit.edu.au>
List: source-changes
Date: 09/21/1999 11:23:35
M Graff writes:
> > > The whole point to the suggestion I made about argument checking was
> > > to catch programmer errors by supplying the file/line and a core dump,
> > > not to ignore errors and continue. We're not linux here. :)
> >
> > So, Core Dumps are for Real Men Who Use BSD? So much for touting
> > ourselves as a robust system...
>
> I believe that hiding errors via return codes and errno settings will
> only create problems. It is better that the application crash as soon
> as such an error is detected rather than try to plod along.
That's not a great solution if NetBSD was used in embedded devices or
other situations where such a crash is not the ideal.
There may be cases in your code that you can't test fully until the
program has been running for a long time. As long as your code checks
the return values of functions, then at least it has a chance to
recover.
As I said, `so much for robust systems'. Anyway, I think we should
agree to disagree on this point, because it's obvious that it's a
controversial one, with people having strong views on both sides.