Peter,
Thanks
On Aug 13, 2012, at 10:43 PM, Peter Brune wrote:
> I fixed it; the prefix fix from last week for PFLOTRAN broke it for composed
> solvers' default options for the line search type, making the richardson
> smoothers run with the basic linesearch and diverge ASAP. The reas
I fixed it; the prefix fix from last week for PFLOTRAN broke it for
composed solvers' default options for the line search type, making the
richardson smoothers run with the basic linesearch and diverge ASAP. The
reason it took so long to show up in dev is due to it being pushed to 3.3
originally,
Something broke; I think I see what it is from your logs. I'll fix it.
- Peter
On Mon, Aug 13, 2012 at 9:17 PM, Barry Smith wrote:
>
> What's up with this stuff? Are they bugs? Bad algorithms? What? It is
> hard to make a major change to PETSc (and use alltests to check that you
> haven't bro
What's up with this stuff? Are they bugs? Bad algorithms? What? It is hard to
make a major change to PETSc (and use alltests to check that you haven't broken
anything) when this crap comes up constantly.
Is there a plan to deal with this or are we just screwed forever?
Barry
1 SN