Fix color future warning

[zoccoler]

This is a small PR that addresses #307 and fixes a couple minor bugs

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 66.66667% with 1 line in your changes missing coverage. Please review.

Project coverage is 79.47%. Comparing base (ba6a572) to head (7cdf261).

Files	Patch %	Lines
napari_clusters_plotter/_plotter.py	66.66%	1 Missing :warning:

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #308   +/-   ##
=======================================
  Coverage   79.46%   79.47%           
=======================================
  Files          16       16           
  Lines        2075     2076    +1     
=======================================
+ Hits         1649     1650    +1     
  Misses        426      426           

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[zoccoler]

If anyone can please give this an OK, I would merge it afterwards, no high-impact changes here

[Cryaaa]

so in general everything LGTM but this only works for environments with napari >= 0.4.19 so there is a backward compatibility breaking change I guess we should mention?

[zoccoler]

True, we would then have to update dependencies as well. I think we are moving in that direction for the next release, right?

[Cryaaa]

That is true but maybe that would create issues with the devbio-napari system... @haesleinhuepf: do you know if relying on a specific napari version would cause issues for the devbio napari install?

[zoccoler]

Let's wait for @haesleinhuepf feedback on this, it can be halted since it is not a major fix.

I would like to eventually have it implemented (maybe it can wait a couple months or so), because it does impact a little my usage of the current plotter by the napari-flim-phasor-plotter plugin

[zoccoler]

I broke this PR into 2.

• here, only changes needed to fix the FutureWarning will be kept
• the other minor changes were moved to #313

This PR should remain open until we are sure we do not break devbio-napari installation.

[ziw-liu]

As of napari 0.5.0 this is now an error:

TypeError: _add_layer_from_data received an unexpected keyword argument ('color') for layer type labels

It would be great if this fix gets into a release soon.

[zoccoler]

OK now this is a bug with napari>=0.5.0.

The workaround until this is fixed is to use a lower napari version...

[jni]

this only works for environments with napari >= 0.4.19 so there is a backward compatibility breaking change

You can check the napari installed version and pass one key or the other based on that, which is what we did with napari-ome-zarr, here:

https://github.com/ome/napari-ome-zarr/pull/112/files#diff-3ac23664593f6255d26806df61d2364f3d3f0b66c53a4fad561a70877093239aR134-R139

Let us know if you have any questions about the new API!

[zoccoler]

You can check the napari installed version and pass one key or the other based on that, which is what we did with napari-ome-zarr, here:

https://github.com/ome/napari-ome-zarr/pull/112/files#diff-3ac23664593f6255d26806df61d2364f3d3f0b66c53a4fad561a70877093239aR134-R139

Oh nice @jni ! I will give this a try! Thanks!

[zoccoler]

With the current state of this PR, there is no need to check for the napari version, it should work whether the Labels layer has the .color attribute or not because the colors dictionary is always converted with DirectLabelColormap (only available with napari>=0.4.18 TMK) into a colormap and assigned to the .colormap attribute.

I tried to install this version on top of the latest devbio-napari version (0.10.1) but I could not make it work when installing with pip becasue it brings npe>=0.7.0,

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
devbio-napari 0.10.1 requires npe2<0.7.0, but you have npe2 0.7.7 which is incompatible.

This seems to be a requirement of napari>=0.5.1. For example, if I try to downgrade npe2 to 0.6.2, I get:

- package napari-base-0.5.1-pyh9208f05_0 requires npe2 >=0.7.6

So, napari-clusters-plotter may still work if installed with mamba/conda along with devbio-napari, but we can only test that after generating a new recipe in conda-forge (TMK).

I suggest we merge changes here regardless of the problems above because otherwise this is a bug of this plugin when napari>=0.5.0. If this new version impacts devbio-napari installation, we should create an issue there and work on a solution from the outside, but not here.

@jo-mueller, if you agree, could you merge this?

[jo-mueller]

Agreed @zoccoler, thanks for the fix!

[jni]

the colors dictionary is always converted with DirectLabelColormap (only available with napari>=0.4.18 TMK) into a colormap

It became available in 0.4.19. We should probably start adding .. versionadded:: directives to docstrings when we add new things... 😅🙏

[zoccoler]

the colors dictionary is always converted with DirectLabelColormap (only available with napari>=0.4.18 TMK) into a colormap

It became available in 0.4.19. We should probably start adding .. versionadded:: directives to docstrings when we add new things... 😅🙏

You are right, my mistake! I had looked into napari API for other versions to check that. DirectLabelColormap only "shows up" in 0.4.19 (https://napari.org/0.4.19/api/index.html), thanks for the correction!

We have 0.4.19 as a requirement here (https://github.com/BiAPoL/napari-clusters-plotter/blob/44a658789fc119e339e320e4cab10238d67781eb/setup.cfg#L52C5-L52C19) so it should all be fine 😌