[petsc-dev] Fwd: [petsc-maint #119133] petsc-dev configure crash

Jed Brown jedbrown at mcs.anl.gov
Wed Jun 6 23:17:53 CDT 2012

On Wed, Jun 6, 2012 at 11:00 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>  Am I missing understanding something. Any option in hg to get hg commands
> inside the subrepo behave like ones on the outside?

Can you make MatView_SeqAIJ look up the stack and figure out that it's a
parallel matrix and print itself appropriately? While you're at it, maybe
you can also print information about the DM and the SNES that it came from.

BuildSystem is still its own repository that you can interact with without
informing petsc. Think of committing from inside BuildSystem meaning that
you explicitly don't want to sync with petsc-dev yet. Committing from the
parent directory indicates that you do want to sync.

I think this will be a rare problem and it's easy enough for the next
person to fix. Certainly no worse than the current system, and the subrepo
model has a bunch of other advantages.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120606/14fb3084/attachment.html>

More information about the petsc-dev mailing list