[petsc-users] compile petsc-dev error

[Jed Brown]

2011/3/25 Gong Ding <gdiso at ustc.edu>

> It seems pastix and parms has function name conflict.
>

Indeed, this is the problem with forgetting to namespace. We can roll a
version with just this symbol namespaced, it would be better if the
developers of these packages would namespace their symbols.


> The function ISCreateGeneral now takes an extra parameter PetscCopyMode.
> Which value is compatible to previous version of ISCreateGeneral?
>

PETSC_COPY_VALUES


>
> MatZeroRows takes two extra Vec. Is it possible to keep the old API
> unchanged but
> use a new API?
>

That becomes an untenable maintenance burden (not so much in this particular
case, but code changes are already required for a few other cases). You can
check the PETSc version and provide a wrapper with the old API that sends
NULL for the last two parameters. A language with keyword arguments would be
better in this regard, but no such language is a presently viable
alternative.


> And what will the version of next petsc release?
> Is it will be 3.2 or 4.0?
>

3.2
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20110325/fdb46f67/attachment.htm>