We're not losing it.  Check out solve().... it still has an 
SNESSetFromOptions() call in it.

As mentioned in the patch Petsc 3.1 breaks when you call SNESSetFromOptions() 
before calling SNESSetFunction() if you are using a matrix free solve.  So that 
one in init() can't happen there fore Petsc 3.1.

We did a bit of testing though... and that call to SNESSetFromOptions in init() 
_is_ important for Petsc 2.3.3.... it appears as if the preconditioning options 
aren't set properly if you only do the SNESSetFromOptions() in solve with Petsc 
2.3.3.... but it works fine that way with Petsc 3.1.

There is still some more investigation to do here to find out exactly when we 
should be calling SNESSetFromOptions() and what the ramifications of not 
calling it early are.... but for now this works.

Derek

On Mar 15, 2011, at 9:53 AM, Roy Stogner wrote:

> 
> It looks as if, as soon as we upgrade to PETSc 3.1, that new #if test
> is going to disable any -snes_whatever options get set on the command
> line.  Am I misreading r4264?
> ---
> Roy
> 
> ------------------------------------------------------------------------------
> Colocation vs. Managed Hosting
> A question and answer guide to determining the best fit
> for your organization - today and in the future.
> http://p.sf.net/sfu/internap-sfd2d
> _______________________________________________
> Libmesh-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/libmesh-devel


------------------------------------------------------------------------------
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to