Codecov Report

Merging #7458 into master will decrease coverage by 4.38%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##           master    #7458      +/-   ##
==========================================
- Coverage   90.08%   85.69%   -4.39%     
==========================================
  Files         454      454              
  Lines       82439    82609     +170     
  Branches    13028    13066      +38     
==========================================
- Hits        74264    70792    -3472     
- Misses       5352     9142    +3790     
+ Partials     2823     2675     -148     

Also the workaround for the scraper (#7416) is not necessary anymore in this PR and should be removed.

Please do remove it here then

All green here and no timeout, I'll proceed with doc building.

Circle was successful too: https://18688-1301584-gh.circle-artifacts.com/0/dev/index.html

This is ready when we need it @agramfort, @larsoner

Tentatively marking for 0.20. It would be good if we could wait for their release, but let's see how long it takes them.

In the meantime @GuillaumeFavelier you have conflicts

@larsoner 0.24 is out since yesterday

I'll fix the conflicts

@GuillaumeFavelier should we merge then?

After giving it more thought, I think it's better to wait for 0.21 except if we really need it for a specific reason before the release.

The version 0.23.1 of pyvista has been tested extensively compared to the latest one.

Testing is always welcome for pyvista 0.24 and (maybe) mne-python 0.21. Let's see how it goes @agramfort 👍

Agreed we should probably wait until 0.21 to advertise PyVista stuff. Having an extra 6 months of testing during which core users/contributors/maintainers, some of whom have asked about 3D viz semi-recently (@bloyl, @drammock) start actually using it in scripts would be beneficial in finding bugs before we force less experienced users to do it.

@GuillaumeFavelier are there latest.inc entries that need to be removed for now until 0.21 is released? Anything else we'd need to remove? Technically people can still do MNE_3D_BACKEND=... but if we don't advertise it I think it's okay to leave it as a "hidden" feature for 0.20.

AFAIK, the only entry about the new _TimeViewer is link_brains:

https://github.com/mne-tools/mne-python/blob/master/doc/changes/latest.inc#L130

Pushed an empty commit to get Circle to build the frontpage examples so we can at least see if one plot works

Thanks @GuillaumeFavelier