Closed ddundo closed 6 months ago
@jwallwork23 I didn't want to do this on your 55_multiple_var_ee branch. I left it to you to delete :)
Thanks. Why did you cancel the CI? Please trigger it so that we can check the tests pass.
Because tests fail anyway since the docker image doesn't have the create_section
fix :)
I've rebuild the Docker image, will trigger tests now.
Hm, there seems to be a problem with assemble_mixed_mass_matrix
after updating. Will look into it.
Hm, there seems to be a problem with
assemble_mixed_mass_matrix
after updating. Will look into it.
Thanks Joe! Everything passed for me yesterday after updating firedrake on Monday, but I updated it again now and get the same tests failing. Sorry about that.
@jwallwork23 @ddundo - I am getting an error with assemble_mixed_mass_matrix
passing the load
instructions to pyop2 without an MPI comm
argument. It looks a test for this was added in pyop2 within the last day might explain why the tests were not failing before and are now.
I think I have it working with a local edit of Firedrake, adding a comm=COMM_SELF
argument to the load
call from assemble_mixed_mass_matrix
. I am not sure this is the correct solution though.
It looks like a PR has been added in Firedrake to address this: https://github.com/firedrakeproject/firedrake/pull/3557
Thanks for looking into this anyway @acse-ej321 :)
@jwallwork23 - It appears the fix for this has been merged into firedrake main
now.
I've rebuild the Docker image, will trigger tests now.
And again. Now fixed. Ready for merge if you're happy @ddundo
Thanks both!
Closes #55.
@acse-ej321 could you please take a look as well? I'm not confident how clear the long comment block is, so I'd appreciate a fresh set of eyes to take a look, since Joe and I are already familiar with the issue :)