sync with PMIX master

[annu13]

No description provided.

[lanl-ompi]

Test FAILed.

[lanl-ompi]

Test FAILed.

[lanl-ompi]

Test FAILed.

[annu13]

Bot: retest

[rhc54]

We should probably remove this debug, or at least replace it with pmix_output_verbose

[annu13]

Will replace with pmix_output_verbose.
On 10/2/15, 5:36 PM, "rhc54" notifications@github.com wrote:

In opal/mca/pmix/pmix1xx/pmix/examples/server.c
#969 (comment):

@@ -343,6 +346,20 @@ static void errhandler(pmix_status_t status,
pmix_output(0, "SERVER: ERRHANDLER CALLED WITH STATUS %d", status);
}

+static void op_callbk(pmix_status_t status,

•                  void *cbdata)
  

  +{
• pmix_output(0, "SERVER: OP CALLBACK CALLED WITH STATUS %d",
  status);
  +}

+static void errhandler_reg_callbk (pmix_status_t status,

•                               int errhandler_ref,
  

•                               void *cbdata)
  

  +{
• pmix_output(0, "SERVER: ERRHANDLER REGISTRATION CALLBACK CALLED
  WITH STATUS %d, ref=%d",
  We should probably remove this debug, or at least replace it with
  pmix_output_verbose
  �
  Reply to this email directly or
  view it on GitHub
  https://github.com/open-mpi/ompi/pull/969/files#r41081057.

[annu13]

@rhc54, please merge this PR. We need to replace pmix_output with pmix_output_verbose in multiple places, I will clean those and submit a PR in pmix repo directly.

[jsquyres]

@rhc54 is probably going to be out for a few days.

Is there a reason you can't update this PR to be correct before we merge it?

[annu13]

It is a trivial change that needs to be done in many places in this file and other files in PMIX. I thought it was easier to do it there and bring to OMPI than the other way round. Anyways I just made the changes and added it to the PR. This is good to be merged once the build is successful.

[jsquyres]

@annu13 I'm sorry; that's what I actually meant: that you would put this stuff on the PMIX master, and then add another "sync" commit over here on this PR. I agree that the direction of code should always flow from PMIX to OMPI -- not the other way around.

Have you committed this stuff on the PMIX master yet?

Also, what's the merge commit on this PR? It doesn't seem like that should be here -- i.e., it seems like that is a problem...?

[annu13]

On 10/6/15, 7:30 AM, "Jeff Squyres" notifications@github.com wrote:

@annu13 https://github.com/annu13 I'm sorry; that's what I actually
meant: that you would put this stuff on the PMIX master, and then add
another "sync" commit over here on this PR. I agree that the direction of
code should
always flow from PMIX to OMPI -- not the other way around.
Have you committed this stuff on the PMIX master yet?
The debug print fix commit
(openmpi@5787e92
37db9) is not in PMIX master. We will fix it in PMIX master bring it in
when we sync next time.
Also, what's the merge commit on this PR? It doesn't seem like that
should be here -- i.e., it seems like that is a problem...?
�
The merge commit corresponds to syncing my fork(annu13/ompi) with
open-mpi/ompi. All the commits were picked up by git automatically. I am
not sure why that commit is causing a failure.
Currently pmix2 is broken on OMPI master, this PR bring the pmix2 fixes
from PMIX master. Could you merge the PR partially ignoring today�s
commits so we can have pmix2 working again.

Reply to this email directly or
view it on GitHub
#969 (comment).

[jsquyres]

@annu13 I'm confused -- I thought you wanted all code to flow from the PMIX tree to OMPI (and not the other way around). Can you commit the 5787e92 patch to PMIX and then re-sync over here to OMPI again? That might also get rid of the (seemingly) unnecessary merge commit...?

[annu13]

fixed debug stmts in pmix master and synced with pmix master (pmix git repo rev git69c398e)