Re: [OctDev] Re: bug octave (Segmentation fault)

[Andy A. <ad...@si...>]

David Bateman wrote:
>
>splu(sparse([1 1], [1 2], [1 1], 3, 3))
>
>also seg faults.... Thus the only real solution is a fix in SuperLU
>itself..

I agree. I have a few other sparse bugs that need fixing.
Essentially, SuperLU does not handly problems gracefully.
Fixing this will require some modifications to SuperLU code

(I've checked the relevant parts of SuperLU into octave-forge
already, because I needed to make some mods to remove
verbose printing)

I'm planning to get to work on this soon, but if anyone else
would like to help, tell me. Also, if someone knows of a
really stable sparse library - it may still make sense to
switch to that.

Andy