[petsc-dev] moab nightlybuild failure

[Satish Balay]

On Fri, 28 Jun 2013, Satish Balay wrote:

> On Fri, 28 Jun 2013, Jed Brown wrote:
> 
> > Satish Balay <balay at mcs.anl.gov> writes:
> > 
> > > And just as with git - were we have to query [or store] the current
> > > local state [so as to compare with whats in moab.py or equivalent] -
> > > we would have to store & query the current state of stuff obtained
> > > from tarballs in a similar manner.
> > 
> > Given an arbitrary working tree, how would you verify whether it's
> > up-to-date with a stated URL, short of always downloading (and caching?)
> > the URL and unpacking it every time?
> > 
> > Compare to
> > 
> >   $ git checkout -f $gitcommit

BTW: you would still have to do extra checks [or stash state] here to
see if you have to rebuild [due to changing working tree] or not.

[or abandon the 'save recompile if possible' feature]

Satish

> 
> Yeah - git checkout is simple at this step. - but it has other burdens
> at other stages [require autoreconf]
> 
> Wrt tarball - I was thinking: - we would cache the url [not the
> tarball] - similar to the way we cache the make/configure-options used
> to rebuild package - and redownload - if no match.
> 
> Satish
>