itaps-parallel "All" convention

[txie at scorec.rpi.edu]

If they are only name changes, I do not have any problem.

Karen, there is one more function that is collective in the specification:

    iMeshP_createGhostEnts()

Maybe it is also need name changing.


Thanks,

Ting




>
> I do not see any major problems from our side (Ting can you also confirm
> this).
>
> One other possibility is to use iMeshP_waitForRequests (I think 'All' can
> be implicit) but anyway will be fine with me...
>
> - Onkar
>
>>
>> In iMeshP, we have used the suffix "All" to indicate that a function
>> must
>> be
>> called collectively; examples include iMeshP_syncPartitionAll and
>> iMeshP_syncMeshAll.
>>
>> We break that convention in two or three functions (at least):
>> -  iMeshP_WaitAll is not collective.
>> -  iMeshP_load is collective
>> -  iMeshP_save would often be collective.
>>
>> So I propose the following changes:
>> -  change iMeshP_WaitAll to iMeshP_waitForAllRequests; change its
>> siblings
>> to iMeshP_waitForRequest, iMeshP_waitForAnyRequest, etc.
>> -  change iMeshP_load to iMeshP_loadAll.
>> -  change iMeshP_save to iMeshP_saveAll.
>>
>> Does anyone mind if I make these changes soon?  I realize they will make
>> our
>> tutorial notes slightly incorrect, but since they are just name changes,
>> they shouldn't set back our development and we could at least have the
>> correct names in the exercises.
>>
>> Karen
>>
>>
>>
>>
>
>
>