The main thing we wanted to discuss was to check in and make sure that the text we have for the I/O is still sufficient and see if there is anything else worth pursuing with respect the FT + I/O. The current behavior is that if there is a failure, the state of the file pointer is undefined and you essentially have to start over. Also, because I/O functions don't synchronize, there's no guarantee of failure notification via the I/O functions and the user might need to provide their own notification mechanism using communicator-based functions. If the I/O group doesn't have anything to add to that, then we could probably push that discussion out or fold it into the I/O time. The only conflict there is the Tools WG and as far as I know, there isn't anyone in FT and Tools. Thanks, Wesley On May 21, 2013, at 9:21 AM, Mohamad Chaarawi <[email protected]> wrote:

Hi All,

Are there any important discussion items about FT and I/O that warrants a full hour of joint WG time at the Forum meeting in 2 weeks? I remember reading here once that there might be something to be done to the I/O chapter, but don't recall any items for discussion. If there is not much to talk about, I would ping Martin to cancel this slot or just merge it into the I/O WG time since I don't have much to discuss there anyway.

Thanks, Mohamad _______________________________________________ mpi3-ft mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft