Re: [ublas] [patch] Fixes for compressed matrices

From: Michael Stevens (mail_at_[hidden])
Date: 2005-03-21 11:30:53

Hi Andreas

On Monday 21 March 2005 11:09, Andreas Klöckner wrote:
> Michael Stevens schrieb:
> > Don't feel disheartened! Although many people are actively involved with
> > uBLAS there are on a couple who can make such commits. It is also always
> > good to have a second opinion on such changes. For myself I was also
> > wondering if the fix should prompt the documentation to be improved with
> > regard to how compressed_matrix is internally manipulated.
>
> In what way do you feel the documentation should be updated? If you'd
> like me to, I'll try and make these changes to make committing the patch
> as much of a no-brainer as possible.

I was thinking of how to describe how the two index arrays are built and the 
invariant that 'complete_index1_data' enforces. That is to say if we keep the 
implementation where index1 may be incomplete we should somewhere document 
it. Do you know if any other compressed matrix libraries yield similarly 
incomplete index1 array?

I have committed the the operation.hpp patch. 

All the best,
        Michael