Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Require Numba 0.57.0+ & NumPy 1.21.0+ #480

Merged
merged 3 commits into from
May 31, 2023

Conversation

jakirkham
Copy link
Member

Align with the rest of RAPIDS on these requirements. Also needed for CUDA 12 support.

@jakirkham jakirkham requested a review from a team as a code owner May 30, 2023 22:51
@github-actions github-actions bot added the conda label May 30, 2023
@jakirkham
Copy link
Member Author

Could someone please add the "improvement" & "non-breaking" labels?

@jakirkham jakirkham requested a review from bdice May 30, 2023 23:27
@AjayThorve AjayThorve added non-breaking Non-breaking change improvement Improvement / enhancement to an existing function labels May 30, 2023
@AjayThorve
Copy link
Member

The docs build is failing, due to an issue where numpy v1.24.3 prevent holoviews to render cuxfilter docs charts, should be fixed in #479 here

@bdice
Copy link
Contributor

bdice commented May 31, 2023

The docs build is failing, due to an issue where numpy v1.24.3 prevent holoviews to render cuxfilter docs charts

@AjayThorve Is there an upstream bug in holoviews / numpy that we can link to in that PR so that we know when this limitation is lifted? This is important to track for all of RAPIDS, because cuxfilter is installed by default in the rapids metapackage and users may have environment conflicts that require NumPy 1.24. We've seen NumPy 1.24 compatibility requests for cudf and were able to lift this upper bound.

@AjayThorve
Copy link
Member

@AjayThorve Is there an upstream bug in holoviews / numpy that we can link to in that PR so that we know when this limitation is lifted? This is important to track for all of RAPIDS, because cuxfilter is installed by default in the rapids metapackage and users may have environment conflicts that require NumPy 1.24. We've seen NumPy 1.24 compatibility requests for cudf and were able to lift this upper bound.

Hey @bdice, so this issue is already fixed in datashader (holoviz/datashader#1176), although it's not yet part of any release(the latest released datashader version is 0.14.4, and this PR is part of the 0.14.5 milestone). How do you suggest we proceed in this scenario?

@jakirkham
Copy link
Member Author

Can we raise an issue on their repo requesting a release (and link that issue here)?

@AjayThorve
Copy link
Member

Have asked for an ETA for release here (https://discourse.holoviz.org/t/module-numpy-has-no-attribute-warnings/5402)

@hoxbro
Copy link

hoxbro commented May 31, 2023

We have released datashader 1.15, and are working on adding it to conda-forge here. So hopefully you will soon be unblocked.

@ianthomas23
Copy link

Conda-forge datashader-feedstock v0.15.0 PR has been merged so you should be able to proceed shortly.

@jakirkham
Copy link
Member Author

jakirkham commented May 31, 2023

Thanks all! 🙏

Copy link
Contributor

@bdice bdice left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also needs to sync with / update dependencies.yaml.

@AjayThorve
Copy link
Member

thanks everyone, #479 should fix the CI issues, and this PR should be good to merge after that

@AjayThorve
Copy link
Member

AjayThorve commented May 31, 2023

@jakirkham can you rebase with default branch, that should fix the CI failures

conda/recipes/cuxfilter/meta.yaml Outdated Show resolved Hide resolved
@jakirkham
Copy link
Member Author

Toggling for CI

@jakirkham jakirkham closed this May 31, 2023
@jakirkham jakirkham reopened this May 31, 2023
Align with the rest of RAPIDS on these requirements. Also needed for
CUDA 12 support.
As NumPy is used internally, list it as an explicit requirement instead
of relying on other dependencies to pull it in.
Copy link
Contributor

@bdice bdice left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please update dependencies.yaml to align. We need numpy and numba pinnings and to make the run section align with the conda list of run dependencies:

cuxfilter/dependencies.yaml

Lines 107 to 128 in 472f839

run:
common:
- output_types: [conda, requirements]
packages:
- bokeh>=2.4.2,<=2.5
- datashader>=0.15
- geopandas>=0.11.0
- holoviews>=1.15.0,<=1.15.4
- jupyter-server-proxy
- libwebp
- nodejs>=14
- packaging
- panel >=0.14.0,<=0.14.1
- pydeck>=0.3,<=0.5.0
- pyppeteer>=0.2.6
- pyproj>=2.4,<=3.4
- output_types: conda
packages:
- cudf=23.06
- cuspatial=23.06
- dask-cuda=23.06
- dask-cudf=23.06

(requested changes previously: #480 (review))

@AjayThorve
Copy link
Member

Please update dependencies.yaml to align. We need numpy and numba pinnings and to make the run section align with the conda list of run dependencies:

@bdice, I pushed the requested changes

@AjayThorve
Copy link
Member

/merge

@rapids-bot rapids-bot bot merged commit 871e0db into rapidsai:branch-23.06 May 31, 2023
@jakirkham jakirkham deleted the req_numba_57 branch May 31, 2023 19:26
@jakirkham
Copy link
Member Author

Thanks all! 🙏

@bdice bdice mentioned this pull request Jul 11, 2023
4 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
conda improvement Improvement / enhancement to an existing function non-breaking Non-breaking change
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants