itaps-parallel New version of iMeshP.h committed

[txie at scorec.rpi.edu]

Karen and Onkar,

Thanks for the discussion. I am sorry for this confusion. I will look into
the FMDB_iMeshP implementation and modify this function in "offset" way
very soon.

Thanks,

Ting



>
> My preference would be to stay closer to other iMesh functions where
> similar  types of array exists (in concept)... I guess others (Carl,
> Tim...) can help me on this one...
>
> If we decide to go with the way FMDB does it (as of now) then I agree with
> Karen to change the name from "offset" to "numNewPerOld" or something...
> good catch Karen... now if I think about it I do not know why FMDB does it
> "numNewPerOld" way if the array is called offset... I will ask Ting to
> look into it...
>
> In summary, if others tell a solution (as per iMesh standards) in a day or
> two then we take that solution... otherwise I will say FMDB needs to be
> updated and use "offset" as offset not "numNewPerOld" because the
> documentation clearly explains offset... the only example getting affected
> by this (that I can think of) is mesh adaptivity which I can take care of.
> This issue I think we will clear before SC08 as I plan to talk about this
> function in tutorial.
>
> Thanks,
> Onkar
>
>>
>> OK; I don't care which way we do it.  If we do it your way, I will
>> change
>> the name of the variable from "offset" to "numNewPerOld" or something
>> like
>> that, since it is more of a count than an offset.
>>
>> Let me know which approach you think is more natural.  Then either I'll
>> change the documentation or you'll change your code.
>>
>> Thanks for your reply.
>> Karen
>>
>>
>>
>> On 11/6/08 7:20 PM, "Onkar Sahni" <osahni at scorec.rpi.edu> wrote:
>>
>>> Karen,
>>>
>>> Thanks for the nice summary.
>>>
>>> See my response below (inlined) on two items.
>>>
>>>> -  How is the offset array in iMeshP_replaceOnPartBdry used?  I
>>>> guessed
>>>> the
>>>> following; Onkar, let me know if it is correct.  If it is correct, do
>>>> we
>>> need offset_size?
>>>>       offset  (In)  Index into new_entities; old_entities[i]
>>>>                     is replaced by new_entities[offset[i]] to
>>>>                     new_entities[offset[i+1]-1].
>>>
>>>   As of now FMDB assumes that old_entities[i] has offset[i] "new ents",
>>> following snippet of code may help:
>>>
>>>   int count = 0;
>>>   for(int i=0; i<old_entities_size; i++) {
>>>     ...
>>>     for(int j=0; j<offset[i]; j++) {
>>>       child[i] = new_entities[count+j];
>>>       ...
>>>     }
>>>     count += offset[i];
>>>   }
>>>
>>>   But I do not see any problem to accomplish the same in the way you
>>> described it (and as it is in recent iMeshP.h). Ting can provide more
>>> details.
>>>
>>>>
>>>> -  Add functions to either pull or accumulate tag data as proposed by
>>> Jason.
>>>>
>>>
>>>    Karen, I see the status of iMeshP_pullTags{Ent} here. Sorry for
>>> sending
>>> you questions separately.
>>>
>>> Thanks,
>>> Onkar
>>>
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>>
>
>
>