Skip to content

Conversation

@nwlambert
Copy link
Member

Just adding a config file to ignore 403 and 405 link errors to see if it fixes those tests.

I also looked at switching to the more supported https://github.com/UmbrellaDocs/action-linkspector but am a bit unsure how to support the matrix option for directories in the config file there. can try if it is worth upgrading.

Seeing if this works to fix failing tests
add config link to nightly yml
@nwlambert
Copy link
Member Author

The link checker passes at least now. seems like with v5 there is a numpy related error popping up? does the test_environment.yml need updating? Also the circuit plotting in some of the qip examples fails, maybe rushi can update those in his PR

Updating numpy and scipy to see if it resolves array error failure
@nwlambert
Copy link
Member Author

nwlambert commented Oct 18, 2024

ok, with some fiddling I can reproduce this locally; the 'python -m pip install jax jax[cpu] equinox diffrax' in notebook_ci.yaml overrides the test_environment definitions, and installs numpy==2.1.2 which doesn't work with the pinned version of matplotlib. better to pin to a lower version of numpy somehow, or update everything in test_environments?

Separate environment yml files for v4 and v5 and updates those for v5 to be something more uptodate, and fix matplotib/numpy clash.
Fixes: matplotlib w_xaxis and axes3d breaking changes,  nmmcsolve() requires the time-dep function to be of type coefficient, brmesolve() parameter order changed.
Small PEP 8 fixes
@nwlambert
Copy link
Member Author

Fixing the bugs revealed more bugs. I think I caught everything, just left the qip plotting errors for rushis PR (there were a few tricky things updating these to the matplotlib render method, so better they are done in that PR).

Fixing the pseudo-probability notebook also revealed that changes in axes3d made many of the plots show up blank. I will make a separate issue to check other notebooks for this.

Copy link
Member

@pmenczel pmenczel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you!

@BoxiLi
Copy link
Member

BoxiLi commented Oct 21, 2024

Thanks! Yes, the qip error is because of the GSoC updates. I think the best way is to replace them with qc.draw("matplotlib"). We can do that in the other PR.

I feel like we should set up a regular weekly test for the tutorial like Eric did for qutip-jax.

@nwlambert
Copy link
Member Author

Thanks! Yes, the qip error is because of the GSoC updates. I think the best way is to replace them with qc.draw("matplotlib"). We can do that in the other PR.

I feel like we should set up a regular weekly test for the tutorial like Eric did for qutip-jax.

thanks boxi. i did try updating, but one of the examples plots a little bit brokenly, somehow the custom latex in a label doesn't get parsed correctly, and another that tries to draw after the gates have been refactored just fails, not sure why. maybe good for rushi to check these edge cases

Add comment on use of config file for link checker, update v5 env to python 3.12
Changing to python 3.12 somehow revealed more pep8 problems.
@nwlambert nwlambert merged commit 65119be into qutip:main Oct 22, 2024
1 of 3 checks passed
@nwlambert nwlambert deleted the fix-broken-links branch October 22, 2024 07:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants