Upgrade for bokeh interactivity and pyvista with trame backend
This is a MR to upgrade/install the necessary libraries to allow bokeh and pyvista to work properly and interactively.
Merge request reports
Activity
I see that now
vtk==9.3.0
is getting installed in this build But according to the comment herevtk==9.2.0
would be preferredEdited by Lia Domide
If we can't make
trame
backend to function, I think we can go only as far aspyvista==0.41.0
From @teodora "in version 0.42.0 pyvista removed deprecated jupyter backends (also pythreejs)"
Also, what version of
ipykernel
we get in this installation (ipykernel==6.29.2
maybe ?)Edited by Lia Domideadded 1 commit
- e962de76 - Update requirements to make pyvista work with trame
Dear @elmath , This is the final version of our merge request that includes all libraries (and their version) which are necessary for tvb-widgets. Please let us know if there are any issues with the libraries/versions we proposed.
Thanks @rominabaila! I am testing this and will let you know :)
@rominabaila @ldomide Starting with this version of the Lab image, all widgets will be installed as part of the ESD and loaded at Lab startup (see ri/tech-hub/platform/esd/ebrains-spack-builds@d73d2f4c, 7e7684bc). I opened a MR to include any missing dependencies, in ri/tech-hub/platform/esd/ebrains-spack-builds!665 (merged), I think just a few are still missing.
Thanks for the update, @elmath !
I see that you included the libraries that we listed in requirements.txt as spack packages, but I have a question regarding
libgl1-mesa-dev
andxvfb
that I added here !43 (diffs): how should these 2 be addressed?perfect. then, let us see the latest form and check if something is still missing there.
BTW: in the new form it is a bit more difficult to test locally, right? in the past we tried a local Dockerimage with the lab + a manual created kernel, now we would need the access to Spack build caches from local, right ?