On Sun, Aug 14, 2011 at 13:11, Barry Smith <bsmith at mcs.anl.gov> wrote:
> Yes, the publisher is required to insure that all memory references that > they publish as fields remain around as long as the AMS memory exists. > This is the problem. There are things that the monitors and viewers compute that are not backed by memory. We still want to be able to see these when observing a computation. So we need to somehow communicate information that is only on the stack to a GUI or database that is monitoring a computation. Since we already want to transmit options, we could create some similar channel (maybe buffered by AMS) or we could just call it a viewer and have a way to hook an AMS client into that viewer (i.e. with a protocol that is not too mismatched). -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110814/b5c1c708/attachment.html>
