[CIG-SHORT] std::bad_alloc() errors when running PyLith in parallel - bug fixed

[Brad Aagaard]

PyLith users,

On Fri I found a bug in PyLith that may have been there for quite some 
time. This caused crashes with the error message reporting a 
std::bad_alloc when running in parallel with faults. I think we have 
seen users report this error on some occasions over the past year, and 
our explanations or suggested actions probably did not fix this problem. 
It appears that the bug may have manifested itself on just a subset of 
platforms, making it more difficult to track down. The fix has been 
pushed to branches/v1.6-stable and the trunk, so users building from the 
repository should update and rebuild both PETSc and PyLith.

We plan to release PyLith v1.6.2 this week and apologize for the delay 
in making these fixes available. This release will fix a number of bugs 
related to fault friction (especially when faults are in tension or have 
zero shear stress) and running in parallel.

Brad