On Wed, 2014-02-26 at 17:14 -0600, Edgar Gabriel wrote:
> that was my fault, I did not follow up the time, got probably side
> tracked by something. Anyway, I suspect that you actually have the
> patch, otherwise the current Open MPI trunk and the 1.7 release series
> would not have the patch after the last ROMIO update - at least I did
> not reapply it, not sure whether Nathan did.
Well, thanks for apologizing for not following up. I was about to get
good and angry, but you've taken away my righteous fury! :>
I've gone through six years worth of OpenMPI romio changes: everything
r16693 | adi | 2007-11-08 06:54:48 -0600 (Thu, 08 Nov 2007) | 2 lines
and picked up a few changes. I'm sure everyone will be happy that I
found the "fix a catamount build warning" change, that was never sent
upstream to us :> . Do you know of any other old patches I should
I've also done a diff of the two source trees. OpenMPI ROMIO is
- some Blue Gene fixes (probably not relevant?)
- some large-datatype processing fixes (probably more interesting).
- better (any) handling of retrying write(2) and read(2) system calls
when they return less than requested
- better memory cleanup in error cases
- a configure-time check for the need for constification of bindings.
(not relevant: openmpi is constified)
- external32 support routines (but external32 is only "mostly there".
Dave knows the details)
I saw mentions in the SVN logs of a pending "updated PanFS driver" but
that was several years ago, and the two drivers are the same right now.
Is there a secret PanFS repository I should raid?
There are still a good deal of Ompi-specific changes that I don't want
to pull into ROMIO, especially anything dealing with the build system.
I should bring this up on openmpi-devel, but I pushed these changes to a