Closed alessandrofelder closed 10 months ago
Needs to be rebased onto #67
Not sure what the rebase has done, but this looks like it's a bug in the unit test itself rather than in the implementation.
I can pip install -e .
this branch, and quite happily be downloading a new atlas whilst looking at an existing one (see below). This is in a clean mamba
environment with python 3.8 (which is the failing case).
Updates:
Argh! I know what's likely happening: my assumption didn't hold on the CI... (and this then breaks the rest of the tests :cry: )
I'll have to do a small refactor
Argh! I know what's likely happening: my assumption didn't hold on the CI... (and this then breaks the rest of the tests 😢 )
I'm able to get the tests to pass w/ Python 3.8, 3.9, and 3.10 via tox locally, so that provides extra evidence for your claim.
Patch coverage: 85.71%
and project coverage change: -0.61%
:warning:
Comparison is base (
adf5ce6
) 97.32% compared to head (3590a0e
) 96.72%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
The line that codeCov is unhappy about is actually covered implicitly, because the test checks the the argument of the emitted signal, which is the return value of the thread worker
Description
What is this PR
Why is this PR needed? Downloading an atlas should not block the napari viewer
What does this PR do? Offloads the atlas downloading to a separate thread.
References
Closes #7
How has this PR been tested?
Existing unit test adapted to wait for worker to finish.
Is this a breaking change?
Nope.
Does this PR require an update to the documentation?
n/a
Checklist: