[mpich2-dev] [mpich-discuss] Overflow in MPI_Aint

[Pavan Balaji]

Joe,

> There are a number of MPI_Aint changes that are not in 1.1.  For 
> example, there is a block of changes in configure.in that relates to 
> this, and it is only in the BGP version.

I'm shifting this part of the discussion to mpich2-dev, since it's more 
developer oriented.

In the top-level configure, the MPIR_Pint stuff should be in 1.1.1. With 
respect to the MPI_OFFSET_TYPE part, is there a need for this? This is 
cross-compile safe, as far as I can tell (there have been some cleanups 
in the trunk, but this part should be correct even in 1.1). Could you 
explain this a bit more?

With respect to the rest of the patch, a lot of it seems to be based on:

1. git-svn differences (e.g., .gitignore files).

2. variables that are automatically filled in by svn (e.g., $Revision$).

3. changes in the test suite.

Apart from these, I do see a few more changes that need to be 
integrated. The dcmfd changes should be easy enough to integrate. I'll 
look into the top-level changes as well.

  -- Pavan

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji