On 03/25/2014 11:22 AM, Dave Love wrote:
Rob Latham <[email protected]> writes:
Yes, I built it, but haven't actually tried running it because the
maintainer reported corruption using OrangeFS instead of older PVFS2,
and I wondered if there was more information; I'll prod him. I guess it
isn't on topic here, assuming it's an openmpi specific problem. Thanks
anyway.
I will gladly incorporate any patches that fix PVFS problems if
someone submits them upstream to ROMIO (me). I've fixed the
ad_pvfs2_io_dtype.c problem, but I don't know aything about this
second patch, and the python script doesn't make sense to me.
There's a patch on the issue tracker:
<https://svn.open-mpi.org/trac/ompi/ticket/4342>. I should have looked
for a ticket.
Ugh, thanks for the link to the ticket. Edgar Gabriel's approach is not
exactly what I would have done (or did do in MPICH). Hope I am not too
late to set things straight.
Also, I didn't read/check vithanousek's reply closely enough. I'd taken
a patch from openmpi 1.7 that changed ADIOI_PVFS2_IReadContig to
ADIOI_FAKE_IreadContig rather than NULL.
setting to NULL seems like a horrible idea.
==rob
--
Rob Latham
Mathematics and Computer Science Division
Argonne National Lab, IL USA
_______________________________________________
Pvfs2-users mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
