write_restart, read_restart issues with 6Dec13 release

No, it is not from a previous version. The in.elastic example creates a structure, runs a minimization, writes a restart files, then uses that restart file as a starting point for future portions of the script. The error comes from reading a restart file written by the same script.

I'm getting the same error with a trivial example, followed by a segfault. I assume there's some discrepancy in the new format between what write_restart writes, and what read_restart is looking for. I'll have a quick look now and see if I can spot it.

I think I've tracked the problem down. Patch attached, there were a couple of issues:

1) Premature close of the file near line 176, leading to the segfault
2) No read of the PROCSPERFILE field just before the peratom section

I haven't tried writing from multiple processors (restart.% format), so there could be other issues waiting there.


read_restart.cpp.diff (733 Bytes)

thanks - I made a couple last-minute logic changes to accommodate the
coming MPIIO options and must not have rerun all my tests.

I posted a 7Dec patch that should fix these problems and another
one that turned up as well.