Fix off-by-one error in Abaqus element renumbering#3699
Fix off-by-one error in Abaqus element renumbering#3699roystgnr merged 3 commits intolibMesh:develfrom
Conversation
|
The This PR should probably reference #3640 (where I introduced the bug) and idaholab/moose#26001 (our first victim of the bug). |
|
And the |
|
Job Coverage on 986b7ef wanted to post the following: Coverage
Warnings
This comment will be updated on new commits. |
||||||||||||||||||||||||||
I translated 1-based-to-0-based for Abaqus-to-libMesh numbering when trying to look up elements for e.g. subdomain ids, but I forgot to translate that when *creating* the elements in the first place.
I love debugging with ASCII files, but even gzip makes these things 5x smaller.
These aren't much, but they would have caught our latest regression at least.
|
This is making me a bit nervous. We added UMAT and UEL support in MOOSE and have some users relying on it. Both plugins do expose the element number, which we test, but unfortunately only with MOOSE generated meshes. Not much we can do about this now, but I'll send an email to our potential users to let them know about this change. Tag @recuero |
Not sure if by "this" you mean the bug (which might be worth a new libMesh submodule update ASAP; that would also get us the Nanoflann update and the fix for clang 15+ FPE issues, neither of which I would mind seeing rushed downstream) or the recent "respect Abaqus numbering" behavior change (which we could make optional if it's causing anyone problems, though I'd still want the old "just number sequentially and ignore the file" behavior to be non-default). |
|
Job Test MOOSE on 986b7ef : invalidated by @roystgnr Possible threaded race condition in misc/solution_invalid.immediately_print_invalid_solution ? Doesn''t look Abaqus related, anyway. |
|
I didn't see that this is a fix for #3640. Carry on then :-) |
3 participants
I translated 1-based-to-0-based for Abaqus-to-libMesh numbering when trying to look up elements for e.g. subdomain ids, but I forgot to translate that when creating the elements in the first place.
I want to add some more test coverage before merging this, but I'm putting it up now to get it available and to make sure there are no side-effects.