Discussion related to cado-nfs

[Paul Zimmermann <Paul.Zimmermann@inria.fr>]

       Dear Robert,

that would be possible, however that would hide potential bugs in the sieving 
code.
Even if warnings are issued for composite factors, I'm sure nobody will worry 
about
them. With errors, the good thing is that people do report the corresponding 
issue,
and the sieving code can be improved.

Best regards,
Paul

> From: Robert Balfour <rhb11931@gmail.com>
> Date: Wed, 9 Feb 2022 18:56:40 -0300
> 
> 
> [1:text/plain Show]
> 
> 
> [2:text/html Hide Save:noname (3kB)]
> 
> Dear all,
> 
> No idea how realistic this is, just something that's been on my mind...
> 
> As we know from issue #30012, sieving with -adjust-strategy 2 very 
> occasionally
> produces a relation with a composite factor. Duplicate removal does not 
> handle
> these well and quits, which is unfortunate because strategy 2 is often 
> faster than
> the default.
> 
> Msieve's duplicate removal simply throws out any relation with a composite 
> factor,
> as well as any other bad relations, and continues smoothly without 
> quitting. As a
> stopgap solution until #30012 is fixed properly (e.g. by merge request 
> #29), would
> it be possible for CADO to handle bad relations in a similar way? This 
> would make
> strategy 2 more practical for jobs small enough that one doesn't want to 
> manually
> intervene.
> 
> Apologies if this would be too difficult to implement - just feels like it 
> ought
> to be easier than a full fix of #30012.
> 
> Best,
> Robert