Re: Abi-MATH works!

Hi Martin,

On Sat, 2005-01-22 at 22:47 +1100, msevior@physics.unimelb.edu.au wrote:
> Nevermind! Rebuilding GtkMathView with your command line arguments above
> gave me a working build.
>
> I get maths again!

that's good news ;-)

> The ideas all worked :-) OK I'll clean things up, move src/af/math/xp into
> a plugin and move the rest of the code into CVS HEAD..

Just as a reminder for ourselves, before the merge we should check that
the tree builds correctly when gtkmathview is _not_ installed and that
the autoconfiguration magic actually works.

I do have a more general question about plugins: there has been
discussion about what plugins should do, but not much about what plugins
should NOT do. In particular, does (will) AbiWord demand for any
specific behavior when, say, a fatal runtime error prevents a plugin
from continuing execution?

I can speak for gtkmathview: there are quite a lot of assertions here
and there in the code, they all use the standard assert macro which
makes the whole application crash in case of an unexpected failure.
AbiWord has a smarter assert macro that fails more gently, also allowing
the user to report the bug if she wishes. So, my questions are:
a) is there a way that AbiWord can capture a plugin crash, however this
occurs, and try some recovery?
b) If not, are plugins supposed to adhere to some conventions so that
their failure does not compromise the whole AbiWord application?

Regards,
--luca