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

Drop {{ pin_compatible('numpy', max_pin='x') }} from Conda recipes #82

Closed
jakirkham opened this issue Jul 17, 2024 · 2 comments
Closed
Labels
improvement Improves an existing functionality

Comments

@jakirkham
Copy link
Member

To prepare for NumPy 2 (specifically for packages building with NumPy), Conda recipes need to update their requirements section to drop {{ pin_compatible('numpy', max_pin='x') }} from run

This is needed as packages built with NumPy 2 remain compatible with NumPy 1. Further conda-forge adds run_exports to numpy 1 and numpy 2 that constrain numpy to ABI compatible versions. More info in upstream issue: conda-forge/conda-forge.github.io#2156

If we have additional constraints to add to numpy (like requiring numpy version 1.23.0 or newer), we can layer this constraint on by adding it to requirements/run

@jakirkham
Copy link
Member Author

Submitted PRs to make these changes

Initially included building with NumPy 2 (and constraining to NumPy 1 at runtime). However due to NumPy constraints in the wheels, this was not possible

So have separated out the pin_compatible changes on its own

Though did leave comments about where NumPy 2 updates need to be made. Should help us catch these when doing the RDFG NumPy 2 migration

rapids-bot bot pushed a commit to rapidsai/cudf that referenced this issue Jul 18, 2024
Part of issue: rapidsai/build-planning#82

Drop `{{ pin_compatible('numpy', max_pin='x') }}` as it is no longer needed. `numpy` has its own `run_exports`, which constraints `numpy` to an API compatible version. More details in issue: https://github.com/orgs/rapidsai/projects/132

So `cudf` now uses that in its recipe builds. Also update `requirements/run` to set the `numpy` lower bound to `1.23` as required by us.

Lastly add todo comments for NumPy 2 update lines.

Authors:
  - https://github.com/jakirkham

Approvers:
  - James Lamb (https://github.com/jameslamb)

URL: #16301
rapids-bot bot pushed a commit to rapidsai/ucxx that referenced this issue Jul 31, 2024
Partially addresses issue: rapidsai/build-planning#82

As UCXX no longer needs NumPy during the build, make sure the Conda packages do not use it when building UCXX.

Authors:
  - https://github.com/jakirkham
  - Mike Sarahan (https://github.com/msarahan)

Approvers:
  - Peter Andreas Entschev (https://github.com/pentschev)
  - Mike Sarahan (https://github.com/msarahan)

URL: #256
rapids-bot bot pushed a commit to rapidsai/cucim that referenced this issue Jul 31, 2024
Partially addresses issue: rapidsai/build-planning#82

Even though cuCIM currently `#include`s `<pybind11/numpy.h>`, the actual C++ code appears not to use NumPy. So this attempts to drop the header and the NumPy build dependency.

Authors:
  - https://github.com/jakirkham

Approvers:
  - Mike Sarahan (https://github.com/msarahan)
  - Gigon Bae (https://github.com/gigony)

URL: #751
@jakirkham
Copy link
Member Author

This is now complete. Closing...

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
improvement Improves an existing functionality
Projects
None yet
Development

No branches or pull requests

1 participant