[Debichem-devel] disabling PeIGS in libglobalarrays-dev

[Jeff Hammond]

I recommend submitted a PR for the Makefile.am fix.  I have no idea why
that is broken, but I don't see it anywhere.

On Wed, Jan 8, 2020 at 6:15 AM Michael Banck <mbanck at debian.org> wrote:

> Hi Drew, hi Jeff,
>
> On Wed, Jan 08, 2020 at 11:18:52AM +0100, Michael Banck wrote:
> > On Wed, Jan 08, 2020 at 12:22:25PM +0800, Drew Parsons wrote:
> > > Hi Michael, the ga build has been missing symbols:
> > >
> > >    ./.libs/libga.a(ga_diag.o): In function `gai_diag_std_':
> > >    ga_diag.F:(.text+0xc65): undefined reference to `fmemreq_'
> > >    ga_diag.F:(.text+0xecd): undefined reference to `pdspev_'
> > >
> > > I raised the issue upstream at
> https://github.com/GlobalArrays/ga/issues/154
> > >
> > > Jeff explained that these are PeIGS symbols, provided by nwchem.  It
> hasn't
> > > been causing problems in practice since nwchem is the only program
> currently
> > > using GA.
> > >
> > > Nevertheless Jeff and his upstream coauthors recommend disabling PeIGS
> in GA
> > > to avoid the problem, especially since scalapack is available for GA
> anyway.
> > >
> > > You enabled PeIGS just in the previous Debian release, so I wanted to
> let
> > > you comment. Should we go ahead and disable peigs, or keep it enabled
> all
> > > the same?
> >
> > Unfortunatey, it's been a year and I don't remember why I enabled it.
> > So I suggest you/we disable it and see whether NWChem still builds/runs
> > fine.
>
> I've uploaded a new GA now which no longer enables peigs. I also fixed
> the test suite which apparentyl has been broken for a long time on
> Debian.
>
> Curiously, I had to patch Makefile.am like this:
>
> https://salsa.debian.org/debichem-team/ga/blob/master/debian/patches/testsuite_log_compiler.patch
> Otherwise, inspecting the *.log files yielded errors like
> 'maybe_mpiexec=mpiexec: command not found'.
>
> Jeff, do you have an idea how this is usually supposed to work? I'm
> happy to file a bug or PR upstream but this seems like a pretty basic
> thing so I am a bit surprised it doesn't work for everybody.
>
> Also, ga++/testing/threadsafecpp.x hang on my notebook and I had to kill
> it :-/
>
> The new results are:
>
> # TOTAL: 78
> # PASS:  65
> # SKIP:  0
> # XFAIL: 1
> # FAIL:  12
> # XPASS: 0
> # ERROR: 0
>
> Let's see what the autobuilders come up with.
>
>
> Michael
>


-- 
Jeff Hammond
jeff.science at gmail.com
http://jeffhammond.github.io/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/debichem-devel/attachments/20200110/2fa06251/attachment-0001.html>