Split into matplotlib and matplotlib-base #178
Conversation
See also the discussion at #2
|
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 ( |
|
Could we use |
recipe/meta.yaml
Outdated
| - mpl_toolkits | ||
| - pylab | ||
| outputs: | ||
| - name: matplotlib-core |
There was a problem hiding this comment.
The first name above is ignored and we are using matplotlib-split just as a convention. That means you need a second output that is named matplolib that depends on matplolib-core and the optional dependencies.
Pinging @msarahan here who knows more about split packages.
There was a problem hiding this comment.
Yes, that's correct. See the numpy recipe for reference: https://github.com/AnacondaRecipes/numpy-feedstock/blob/master/recipe/meta.yaml
There was a problem hiding this comment.
Hm, but running conda build locally definitely produced two packages, and they even worked as expected when I installed them into test environments. I used the implicit metapackages feature. Or do I misunderstand something?
There was a problem hiding this comment.
If it works, you're fine. That has been fragile in the past, but hopefully we have it tested well enough now.
There was a problem hiding this comment.
Ok, thanks, then I’ll leave it for now.
|
I like |
|
Thanks @marcelm. |
marcelm
changed the title
|
Here comes a friendly reminder that this PR is still open. |
|
To avoid naming future naming issues with |
|
Thanks @jjhelmus! |
|
Yay, thanks so much! |
Thank you for the PR! This was "in the works" since #2 but b/c |
The matplotlib package has been split into a matplotlib and matplotlib-base package a while ago, see conda-forge/matplotlib-feedstock#178 . The only difference is that matplotlib-base does not include on pyqt, reducing installation size significantly. Unless matplotlib’s PyQt backend is explicitly required, depending on matplotlib-base is sufficient.
Merge PR #19985, commits were: * Depend on matplotlib-base instead of matplotlib The matplotlib package has been split into a matplotlib and matplotlib-base package a while ago, see conda-forge/matplotlib-feedstock#178 . The only difference is that matplotlib-base does not include on pyqt, reducing installation size significantly. Unless matplotlib’s PyQt backend is explicitly required, depending on matplotlib-base is sufficient.
This closes #2 by splitting matplotlib into a
matplotliband amatplotlib-corepackage:matplotlibbecomes a metapackage that depends onmatplotlib-coreandpyqtmatplotlib-coreis the same as the previous matplotlib package, except that it does not depend onpyqt.Due to the new backend fallback mechanism in Matplotlib 3.0.0, a user who choses to install
matplotlib-corewill get the TkAgg backend automatically.Closes #2. Supercedes #157.
I have had little experience with multiple outputs and it took me a while to get it to work, so I’ll appreciate feedback. In particular, I am not sure whether it was necessary to rename
build.sh. The problem was that it was called even when packaging thematplotlib-coresubpackage, which failed because the script assumes the build dependencies are available.Checklist
0(if the version changed)conda-smithy(Use the phrase@conda-forge-admin, please rerenderin a comment in this PR for automated rerendering)