On Fri, 2017-09-29 at 01:47 +0800, Drew Parsons wrote: > On Thu, 2017-09-28 at 09:56 -0700, Dima Kogan wrote: > > Drew Parsons <[email protected]> writes: > > > > > Heads up on sundials, PETSc 3.8 has just been released. I'll > > > upload > > > to > > > experimental soon, and to unstable as soon as practicable. > > > Hopefully > > > sundials won't have trouble with the new petsc. > > > > I'll try it out when you make the upload. Was there an ABI update > > in > > PETSc? I.e. would a binary re-upload for sundials need to happen? > > It definitely has an ABI update. I think the API is safe (backwards > compatible) so hopefully a binNMU (automatic rebuild on the buildds) > is > all that sundials will need.
Actually a handful of PETSc 3.8 API changes reported at https://www.mcs.anl.gov/petsc/documentation/changes/38.html For the most part just renaming a few existing functions, so should be easy to adapt to if needed. If there's trouble then it might be worth asking sundials upstream about readiness for PETSc 3.8 (I won't drop petsc 3.8 from experimental to unstable until we have some certainty). In the meantime I've now uploaded PETSc 3.7.7, the last point release before 3.8. Same ABI (and API) as other 3.7 versions, just a routine update. Drew
