Solutions sometimes depend on Workset Size about albany HOT 5 CLOSED

Fixed bug in Enthalpy test 20c896
Now test result is independent of workset size. As I expected the bug was very specific of an evaluator exclusively used by the Enthalpy solver.

I'm closing this issue an refer to #49 for the ATO part.

from albany.

#49 is part of this

from albany.

I'm pretty confident that the issue in commit 9116dba is not related to #48 or #49. I think it has to do with an evaluator that is only used in the Enthalpy test, and that I will fix when I'll go back working on that capability. That's why I did not open an issue.

from albany.

true, they are likely unrelated multiple bugs.

from albany.

Your guess on the issue being related to the "last" workset makes sense to me. I will check that. My first guess for the Enthalpy test, is that I'm doing a (tricky) integral along a vertical column, and that I might be wrongly assuming that all the cells in a column belong to the same workset, although that's not guaranteed and it depends anyway on the workset size. Anyway, I'm glad that the change in the workset size exposed this serious bug. I'm not that worried because this is a new/experimental capability that has not been used yet for real applications.

from albany.