Fix/988/pdosworkchain #993
Conversation
mbercx
force-pushed
the
fix/988/pdosworkchain
branch
from
8bb81e0
to
1c6aa3d
Compare
mbercx
force-pushed
the
fix/988/pdosworkchain
branch
from
1c6aa3d
to
68bd925
Compare
| query = QueryBuilder().append( | ||
| InstalledCode, | ||
| filters={'label': label}, | ||
| ) | ||
| try: | ||
| return load_code(label=label) | ||
| except exceptions.NotExistent: | ||
| return query.first()[0] | ||
| except TypeError: |
There was a problem hiding this comment.
Why was this changed? @mbercx I don't understand this nor do I think it is correct. I would revert this
There was a problem hiding this comment.
I'm not 100% what was happening under the pytest hood, but when using pytest.mark.parametrize to test the two possible NSCF output parameters, it seems to somehow have created two instances of the quantumespresso.pw code, and the second test fails with:
aiida.common.exceptions.MultipleObjectsError: multiple Code entries found with LABEL<test.quantumespresso.pw>
My solution is a bit of a late-night hack, I'll admit. ^^
Looking into this further, it actually seems that VSCode running the test in parallel causes the issue. After reverting reverting the change here, the tests still pass fine. Running pytest manually locally also works without issue.
Happy to open a PR to revert, but maybe there is also a more elegant solution to have the tests run properly in parallel?
There was a problem hiding this comment.
Well, the tests are not designed to be run in parallel, especially not if they use the same storage backend. Even if you "fix" this case, there is bound to be a shitload of other problems lying in wait when you have multiple tests run against the same database. So I really think you just shouldn't run them in parallel.
Fixes #988