Add `napari-base` output and reorganize dependencies #63

psobolewskiPhD · 2024-07-13T23:02:14Z

Closes #62

Checklist

Used a personal fork of the feedstock to propose changes
Bumped the build number (if the version is unchanged)
Reset the build number to 0 (if the version changed)
Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
Ensured the license file is being packaged.

Need a repo patch too I think?

Edit by @jaimergp:

I've added a new output, napari-base, following conversation in this PR, this Zulip thread, and this other one.

With this PR we will have these outputs:

napari-base: the core library, with only the required dependencies. Use it for your Pyhton imports but no GUI action.
napari: the application sans the Qt backend. It contains optional yet recommended dependencies for (performant) GUI usage, like the plugin manager or numba.
napari-menu: just the menuinst JSON file, in case you want a desktop shortcut.

Most end-users will want to conda install napari pyqt or conda install napari pyside2.

…pendencies

conda-forge-webservices · 2024-07-13T23:02:23Z

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

psobolewskiPhD · 2024-07-13T23:03:33Z

@conda-forge-admin, please rerender

github-actions · 2024-07-13T23:04:52Z

Hi! This is the friendly automated conda-forge-webservice.

I tried to rerender for you, but it looks like there was nothing to do.

This message was generated by GitHub actions workflow run https://github.com/conda-forge/napari-feedstock/actions/runs/9923450902.

jaimergp

IIRC, this is intentionally left out of the dependencies. It's part of run_constrained so we can still control which version is compatible, but we didn't want it by default. I don't remember the exact wording, but @jni stated that "it'd be crazy if matplotlib had its own plugin manager shipped with the library, so we shouldn't do that either".

psobolewskiPhD · 2024-07-15T18:52:01Z

@jni can you clarify?
I had just advertised how in 0.5.0 plugins could be installed from conda when napari is from conda -- which is how we recommend installing on napari.org
Not being able to do so, without any deprecation, documentation or info to Plugin authors seems like a serious regression.
My understanding of devolving the plugin manager into a plugin was to make it easier to update it and maintain, plus have it be able to be used by others. Not that it would be removed from napari period.

jni · 2024-07-16T02:28:02Z

I'm afraid the wording quoted by @jaimergp is accurate. 😜

The resolution here is down to the question of do we make conda install napari install the whole kit and caboodle and conda install napari-base installs something more barebones, or do we make conda install napari-kit-and-caboodle install everything and conda install napari install the barebones. Personally, I prefer the latter, but I can be pushed to the former with little effort. Either way, I'd love to solve both things in one step. @jaimergp can you point us to examples of multi-recipe feedstocks?

Also, sorry, I merged #64 before seeing this, so the build number needs another bump here regardless.

psobolewskiPhD · 2024-07-16T02:43:19Z

Here's my thoughts:
Conda-forge is the install path of least resistance and what we recommend in docs. IMO it should absolutely be napari[all] and not napari-lite.
I wish we didn't have to specify Qt backed, but that's in the docs -- still many not aware of that change though, but the error is instructive. Meanwhile, in this case, a major feature is just poof, gone. With no easy obvious alternative -- there is no error or anything.
To me this is a massive regression. A more sophisticated user can install however they want -- from git for all I care -- but the GUI functionality is for new users or people just getting into the ecosystem. Especially now that the conda install could install plugins from conda which solves so many issues!
😢

jaimergp · 2024-07-16T08:26:53Z

The resolution here is down to the question of do we make conda install napari install the whole kit and caboodle and conda install napari-base installs something more barebones, or do we make conda install napari-kit-and-caboodle install everything and conda install napari install the barebones.

We've taken steps to make napari more lightweight, so I'd rather stay in that direction (see #56 and specially #47 (comment)). That said, we can do that later and still have napari ship napari-plugin-manager for now. It's not that heavy (yet).

…atch-1

jni · 2024-07-16T09:04:19Z

Some further discussion in Zulip here and here.

The decision for now is to keep the plugin manager in napari and make the napari recipe more feature-complete, while creating a lightweight napari-base package on which napari would eventually depend.

conda-forge-webservices · 2024-07-16T09:13:50Z

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe:

Jinja2 variable references are suggested to take a {{<one space><variable name><one space>}} form. See lines [118].

conda-forge-webservices · 2024-07-16T09:18:13Z

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

jaimergp · 2024-07-16T09:18:37Z

recipe/meta.yaml

@@ -49,7 +49,6 @@ outputs:
        - magicgui >=0.7.0
        - napari-console >=0.0.9
        - napari-plugin-engine >=0.1.9
-        - napari-svg >=0.1.8


Moved to napari

jaimergp · 2024-07-16T09:18:59Z

recipe/meta.yaml


-        # additional dependencies for convenience in conda-forge
-        - fsspec
-        - pooch >=1.3.0
-        - zarr >=2.12.0
-
      run_constrained:


Moved to napari

jaimergp · 2024-07-16T09:19:08Z

recipe/meta.yaml

      run_constrained:
-        - napari-plugin-manager >=0.1.0a2,<0.3.0a0


Moved to napari as a full dependency

jaimergp · 2024-07-16T09:19:36Z

recipe/meta.yaml

        - pyside2 >=5.13.2,!=5.15.0
        - pyside6 <6.5|>=6.7
        - pyqt >=5.12.3,!=5.15.0,<6.0a0|>=6.5,!=6.6.1
-        # Nice for performance, but optional. Inclusion under discussion.
-        # - py-triangle
-        # - numba >=0.57.1  # https://github.com/napari/napari/pull/6467#issuecomment-1819038165


Moved to napari as full dependency.

jaimergp · 2024-07-16T09:20:05Z

recipe/meta.yaml

+        - napari-svg >=0.1.8
+        - numba >=0.57.1
+        - pooch >=1.3.0
+        # - py-triangle  # a bit fragile with bad data, leave out for now


Suggested change

# - py-triangle # a bit fragile with bad data, leave out for now

# - py-triangle # a bit fragile with bad data, leave out for now

Context about fragility: https://napari.zulipchat.com/#narrow/stream/212875-general/topic/napari.20plugin.20manager.20in.20conda.20recipe/near/451712796

More context in napari/napari#6654

Oh, thanks, I'll add this as the comment instead

jaimergp · 2024-07-16T09:20:28Z

@conda-forge-admin, please rerender

…nda-forge-pinning 2024.07.16.07.04.17

jni · 2024-07-16T09:22:52Z

recipe/meta.yaml

+        - zarr >=2.12.0
+    test:  # just to check that the environment can indeed solve
+      requires:
+        - pyside2  # use this opportunity to test solvability with other backend


Does this mean we are testing with pyside2 but not with pyside6 or pyqt or...?

We tested installability with pyqt in the output above (napari-base). In here we just pick pyside2 because we can now, given the new output. We are not running the test suite here, just installability checks and minimum "this thing runs", import wise.

Great, thanks!

jaimergp · 2024-07-16T09:27:24Z

cc @hmaarrfk, because in the past you've been interested in a lighter napari install tree.

hmaarrfk · 2024-07-16T11:34:06Z

recipe/meta.yaml

      run_constrained:
-        - napari-plugin-manager >=0.1.0a2,<0.3.0a0
        - pyside2 >=5.13.2,!=5.15.0
        - pyside6 <6.5|>=6.7


Thanks for keeping things flexible. I don't really understand the plugin manager but cool!!!

hmaarrfk · 2024-07-16T12:04:32Z

My main question is about the plugin engine actually.

Does it leverage conda-forge??? Or will it force the installation of these plugins through pip? Installing through pip in my mind would hurt users in the long run

jaimergp · 2024-07-16T12:11:21Z

Yes, if napari was installed with conda.

hmaarrfk · 2024-07-16T12:17:01Z

great!!! awesome!

psobolewskiPhD · 2024-07-17T01:06:35Z

Thanks @jaimergp

# References and relevant issues  # Description    Describes the actions taken at conda-forge/napari-feedstock#63  --------- Co-authored-by: Juan Nunez-Iglesias <jni@fastmail.com>

Update meta.yaml to include napari-plugin-manager >=0.1.0a2 in run de…

5b2a201

…pendencies

psobolewskiPhD requested review from AhmetCanSolak, goanpeca, jaimergp, jni, royerloic and sofroniewn as code owners July 13, 2024 23:02

update build number

379bfd6

jaimergp requested changes Jul 15, 2024

View reviewed changes

Merge branch 'main' of github.com:conda-forge/napari-feedstock into p…

3913dea

…atch-1

Add napari-base output, redistribute dependencies

379119b

jaimergp changed the title ~~Include napari-plugin-manager >=0.1.0a2 in run dependencies~~ Add napari-base output and reorganize dependencies Jul 16, 2024

lint

5501d9f

jaimergp reviewed Jul 16, 2024

View reviewed changes

MNT: Re-rendered with conda-build 24.5.1, conda-smithy 3.37.0, and co…

3504f2b

…nda-forge-pinning 2024.07.16.07.04.17

jni reviewed Jul 16, 2024

View reviewed changes

jni approved these changes Jul 16, 2024

View reviewed changes

update py-triangle comment

62c929f

hmaarrfk reviewed Jul 16, 2024

View reviewed changes

jaimergp approved these changes Jul 16, 2024

View reviewed changes

jaimergp mentioned this pull request Jul 16, 2024

Add docs about the new napari-base structure napari/docs#457

Merged

jni merged commit e9e6f6a into conda-forge:main Jul 16, 2024
3 checks passed

jaimergp mentioned this pull request Jul 16, 2024

Sync recipe with conda-forge#73 napari/packaging#169

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add `napari-base` output and reorganize dependencies #63

Add `napari-base` output and reorganize dependencies #63

psobolewskiPhD commented Jul 13, 2024 •

edited by jni

Loading

conda-forge-webservices bot commented Jul 13, 2024

psobolewskiPhD commented Jul 13, 2024

github-actions bot commented Jul 13, 2024

jaimergp left a comment •

edited

Loading

psobolewskiPhD commented Jul 15, 2024 •

edited

Loading

jni commented Jul 16, 2024

psobolewskiPhD commented Jul 16, 2024 •

edited

Loading

jaimergp commented Jul 16, 2024

jni commented Jul 16, 2024

conda-forge-webservices bot commented Jul 16, 2024

conda-forge-webservices bot commented Jul 16, 2024

jaimergp Jul 16, 2024

jaimergp Jul 16, 2024

jaimergp Jul 16, 2024 •

edited

Loading

jaimergp Jul 16, 2024

jaimergp Jul 16, 2024

jni Jul 16, 2024

jaimergp Jul 16, 2024

jaimergp commented Jul 16, 2024 •

edited

Loading

jni Jul 16, 2024

jaimergp Jul 16, 2024

jni Jul 16, 2024

jaimergp commented Jul 16, 2024

hmaarrfk Jul 16, 2024

hmaarrfk Jul 16, 2024

hmaarrfk commented Jul 16, 2024

jaimergp commented Jul 16, 2024

hmaarrfk commented Jul 16, 2024

psobolewskiPhD commented Jul 17, 2024

	# - py-triangle # a bit fragile with bad data, leave out for now
	# - py-triangle # a bit fragile with bad data, leave out for now

Add napari-base output and reorganize dependencies #63

Add napari-base output and reorganize dependencies #63

Conversation

psobolewskiPhD commented Jul 13, 2024 • edited by jni Loading

conda-forge-webservices bot commented Jul 13, 2024

psobolewskiPhD commented Jul 13, 2024

github-actions bot commented Jul 13, 2024

jaimergp left a comment • edited Loading

Choose a reason for hiding this comment

psobolewskiPhD commented Jul 15, 2024 • edited Loading

jni commented Jul 16, 2024

psobolewskiPhD commented Jul 16, 2024 • edited Loading

jaimergp commented Jul 16, 2024

jni commented Jul 16, 2024

conda-forge-webservices bot commented Jul 16, 2024

conda-forge-webservices bot commented Jul 16, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jaimergp Jul 16, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jaimergp commented Jul 16, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jaimergp commented Jul 16, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hmaarrfk commented Jul 16, 2024

jaimergp commented Jul 16, 2024

hmaarrfk commented Jul 16, 2024

psobolewskiPhD commented Jul 17, 2024

Add `napari-base` output and reorganize dependencies #63

Add `napari-base` output and reorganize dependencies #63

psobolewskiPhD commented Jul 13, 2024 •

edited by jni

Loading

jaimergp left a comment •

edited

Loading

psobolewskiPhD commented Jul 15, 2024 •

edited

Loading

psobolewskiPhD commented Jul 16, 2024 •

edited

Loading

jaimergp Jul 16, 2024 •

edited

Loading

jaimergp commented Jul 16, 2024 •

edited

Loading