Update env #723
Update env #723
Conversation
|
6 tests related to |
|
Will look into it. Regarding the gridded data, no strong reason but that won't be the case when working on 3D as the Z position of the hits is not gridded. |
gonzaponte
left a comment
gonzaponte
left a comment
There was a problem hiding this comment.
We should have an automated test for this thing. Since we don't we should have a non-automated one at least. Try to run the script both installing and not installing conda. neutrinos would be a good test bench for this as many people will use it there.
mmkekic
force-pushed
the
update-env
branch
3 times, most recently
from
416ac4a to
8f95580
Compare
gonzaponte
left a comment
gonzaponte
left a comment
There was a problem hiding this comment.
This PR updates our packages to a recent version, much needed after freezing them for 1.5 years.
There are some caveats:
- Some tests have been skipped due to a change in behavior of scipy.interpolate that doesn't affect the final result. These are to be fixed soon.
- Some tests have moved from a single-input to a hypothesis-generated input, which may increase test-execution time (we haven't evaluated it).
Otherwise, this is ready to be merged. Good job.
|
Please rebase and merge. |
The packages compatible with python 3.8 as well
In future hypothesis version this will be a keyword-only argument
The examples were drawn using .example that is only for interactive purpuses
pytest will always raise a warning for custom non-documented marks
The scipy function is giving different result. This issue has to be addressed in the future and the tests recovered.
Using non-Enums in containment checks will raise TypeError in Python 3.8
#733 [author: Aretno] When updating some packages in #723 there were some test that were marked as skipped after checking final result did not change performance-wise. This PR aims to update the values of the test so they can be enabled again. [reviewer: gonzaponte] Scipy interpolate changed behavior in the recent package upgrade. The exact methodology change hasn't been found in the changelog, which is annoying. The differences in the values obtained in the tests vary around the percent level, but the overall performance of the algorithm hasn't been affected, according to @Aretno. Thus, the update simply relies on freezing these new values, hoping that they will not change again in a future release of scipy.
4 participants
This PR updates the environment packages.