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.

Just to be sure. Does this limit to numpy < 2?

It should as we don't support 2.0 yet (and might segfault because it calls the wrong ABI), we restricted the version this release.

Just to be sure. Does this limit to numpy < 2?

It should as we don't support 2.0 yet (and might segfault because it calls the wrong ABI), we restricted the version this release.

No, the numpy pin is open.

No, the numpy pin is open.

Let's change it then to numpy >=1.16.0,<2? 😄

One of the builds failed with

error occurred: Failed to find tool. Is `ml64.exe` installed?

I'd be glad if someone more knowledgable about the (windows) build system could chime in here. I'll try to do some digging, but my best guess is that this is a missing dependency for the windows build path...

There is

build_platform:
  linux_aarch64: linux_64
  osx_arm64: osx_64
  win_64: linux_64

in conda-forge.yanl

The cross-compilation for win_64 is probably deliberate, maybe this issue about a regression in VisualStudio is related.

And also this file specifies clang version 17 as compiler, but I see that version 18 has already been released.

I also found this PR which seems to address a similar problem with cross-compilation from linux-64 to win-64. And finally this issue in rust-lang/cc-rs.

I'd be glad for any suggestions here.

@0xbe7a can you help with the Windows build? Linux aarch64 also seems to timeout :/

@0xbe7a can you help with the Windows build? Linux aarch64 also seems to timeout :/

I think the Windows build should be fixed, but I'm not sure yet what's happening with the aarch64-linux builds. It seems like it's not actually timing out (we have a 6-hour timeout for Azure Pipelines) but rather starving the worker process (?).

The aarch64 builds look like an out-of-memory during linking.

Thanks a lot everyone, especially @0xbe7a, for all the help and for fixing the build. Once the reviews are in I'd be happy to merge this.

FYI. Next Polars release we will have numpy 2.0 support. Just to keep you guys busy. ;)

FYI. Next Polars release we will have numpy 2.0 support. Just to keep you guys busy. ;)

Looking forward to getting full numpy-2.0 support. Great job on the 1.0.0 release 👍 .

Hopefully, the next release will just need to unpin numpy and be ready to merge after a single round of CI, then.

Thanks a lot @0xbe7a for updating and merging!