-
Notifications
You must be signed in to change notification settings - Fork 239
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Switch reference tester to deal.II 9.5 #5498
Switch reference tester to deal.II 9.5 #5498
Conversation
e5f8c3f
to
978cd83
Compare
I am chasing a bug in the zero_matrix test that I cannot seem to figure out right now. I am partly writing this as notes to myself (for when I can get back to this PR), but if anyone has ideas they would be appreciated. What I noticed:
So I guess my question is: If this is the case, why did this test work previously? Did deal.II change how constraints work internally? Or was our test broken in some sense? I can update the test to work correctly if I just remove the boundary conditions for temperature, which will correctly lead to a matrix with only zero entries, which will then also trigger the Assert. Would this be the best way to handle this? |
Interesting problem. I was not aware that something inside constraints has changed with 9.5. I do think it is correct/acceptable to have nonzero entries on the diagonal for constrained entries. I will take a look soon, unless @bangerth has an idea. |
978cd83
to
60545f4
Compare
For now I have disabled the prescribed bondary conditions in the failing test I also looked through all remaining tests, and while there are some changes they all look like typical changes that could come with a new deal.II version. I think this is ready to move forward. The one remaining issue is the failing tester for deal.II master, but I think that is unrelated as well, and likely caused by dealii/dealii#16292. It should not be a problem for moving forward with this PR. |
/rebuild |
We've done a lot of work on |
60545f4
to
f5cdf82
Compare
f5cdf82
to
fde90f2
Compare
I revisited this PR, rebased to the current main branch and reran the test locally. If the online testers succeed, I would like to move forward with this PR, because it blocks a number of other open PRs (#5367, #5383) that can be relevant for the user meeting (we should explain we are going to change the list of dependencies). |
The dealii-master check of course fails, and the OSX tester is down. The rest seems to work, so I'll merge. |
We need to switch our reference tester to deal.II 9.5 to move forward with some other PRs (e.g. #5367, #5383). This is the first step. I still need to update the test results after the tester failed.