[mpich2-dev] about (closed) ticket #449 and request access to track

[Dave Goodell]

On Mar 11, 2009, at 4:28 PM, Lisandro Dalcin wrote:

> First of all, Any chance I could make comments on MPICH2's Trac?

Unfortunately, there's no way at the moment for you to be able to get  
a trac account and we get too much spam when we permit anonymous  
posting.  Our IT division here is working on a solution for this but  
it will be a little while still before it's deployed.

In the mean time, you can reply to the "mpich2-maint@" email that you  
received and that should add your comment to the ticket.

> Second, about my recent (and now closed) ticket #449 ...
>
> My patch is the only robust way I can imagine of doing things...
>
> 1) MPI guarantees that when COMM_SELF's free keyval callbacks are
> called, the MPI environment is intact. So we should be able to call
> MPI_Keyval_free() on cleanup callbacks.
>
> 2) Eager freeing (after Attr_put) wil NOT work ... I've already tried
> this ;-) ... MPICH2 seems to reuse keyval values... If you free() a
> keyval, and next create() other... you the new keyval value is the
> equal to the former... Try the code below.

Yes, I believe that (1) is what we also decided was the right fix.   
(2) is clearly wrong per the MPI Standard.

> IMHO, you closed my ticket too fast ... ;-)

I apologize for closing the ticket on you too quickly.  I thought that  
Rob Latham was actually working on this particular issue at this very  
moment, but it turns out I was partially incorrect.  Your ticket was  
reopened as soon as I learned that.

> Moreover, take into account that I've really tested the patch before
> posting it :-) ...

Unfortunately I didn't realize that you had actually attached a  
patch.  Rob and I will look at this and see if we can apply it.

-Dave