On Thu, 19 Jan 2012, Matthew Knepley wrote: > On Thu, Jan 19, 2012 at 10:03 AM, Satish Balay <balay at mcs.anl.gov> wrote: > > > On Thu, 19 Jan 2012, Sean Farley wrote: > > > > > > > > > > HDF5 in their infinite wisdom run tests in the same make target as the > > > > build. One of the Fortran > > > > tests breaks on the Mac. The only solution that worked for me was to > > > > disable that test with a > > > > patch for the source. I mailed them this problem with no response. Feel > > > > free to fix this in a way > > > > that satisfies you aesthetically. > > > > > > > > > > I run version 1.8.8 and haven't had anything blow up when making tests. > > > > For one - if we have to manitain patches - we do them in a repo - for eg: > > at http://petsc.cs.iit.edu/petsc/externalpackages/hdf5-1.8.4/ [and spin > > our patched tarball] > > > > I do not see that being in any way superior, and my way allows us to pull > directly from the HDF5 site.
1. patches are easier to maintain and track in hg. 2. From current experience direct link to external sites is not necessarily superior. [still preferable - but not really superior]. More points of failure [when compared to a single download site]. Also there are python problems with https:// sites with proxy Perhaps I should switch over these [https] tarballs to ftp.mcs.. Satish > > > > If 1.8.8 works without patches - then we could upgrade to that.. > > > If it works, I will not argue. However, if it fails on my machine, I do not > want to change. > > Matt > > > > > > Satish > > > > > >
