Make a conda release of v0.8.x

[Radonirinaunimi]

Would it be possible to also have the conda release of v0.8.0? This would be needed for various tests in NNPDF (see for example NNPDF/nnpdf#2110).

[cschwan]

Yes, we'd need to fix conda-forge/pineappl-feedstock#40 first to make that happen, but so far I haven't had the time to look into it.

[Radonirinaunimi]

Btw, @cschwan if you need someone to experiment with the macos feedstock please let me know.

[cschwan]

Yes, I'm not sure why the installation of the wheels breaks on macOS: https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=980538&view=logs&j=9c5ef928-2cd6-52e5-dbe6-9d173a7d951b&t=20c71c51-4b27-578b-485d-06ade2de1d00&l=1628. The corresponding line in the build script should be: https://github.com/conda-forge/pineappl-feedstock/blob/ac754fd6f8683acc0b4e4b7500923d2d7ce4b2e1/recipe/build.sh#L10.

[Radonirinaunimi]

Thanks, I will have a look.

[Radonirinaunimi]

I can both build the wheel and install it locally so I am not sure yet what is the issue.

The only difference between what is in the remote and what I have locally is that even though I am also using MacOsX_13, given that my python (through conda) was compiled with an older platform the wheel has the following reference: pineappl-0.8.1-cp312-cp312-macosx_10_12_x86_64.whl. But I don't think this could make a difference.

One thing that I don't understand is that it seems that it is trying to install the wheel with a different python version from the (c)python from which the wheel was build: https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=980538&view=logs&j=a8dee1cf-d17f-5cb0-b53f-4a93dfdbfb0c&t=d9dbf137-9a7f-5666-c60c-9e1b2fc31445&l=1616

I will still keep digging.

[cschwan]

@Radonirinaunimi I added you as a code owner to the conda-forge repository, so you should have write access now. Let me know if it doesn't work!

[Radonirinaunimi]

@Radonirinaunimi I added you as a code owner to the conda-forge repository, so you should have write access now. Let me know if it doesn't work!

Perfect! Thanks @cschwan.

[cschwan]

@Radonirinaunimi I released 0.8.2 yesterday, so you can either abandon 0.8.1 and proceed with 0.8.2 (merging 0.8.2 will close the PR for 0.8.1) or you continue with 0.8.1 and repeat your changes with 0.8.2 after you've merged 0.8.1.

[Radonirinaunimi]

@Radonirinaunimi I released 0.8.2 yesterday, so you can either abandon 0.8.1 and proceed with 0.8.2 (merging 0.8.2 will close the PR for 0.8.1) or you continue with 0.8.1 and repeat your changes with 0.8.2 after you've merged 0.8.1.

Yes, I am continuing directly with 0.8.2.

Some of the MacOS platforms are now working (status) but there are still some failures.

[Radonirinaunimi]

The remaining error:

tar: Error opening archive: Unrecognized archive format

seems to be due to the .tar.gz not being downloaded properly. By re-running the actions, the failed ones now pass.

I am not sure what could be done here as this is really unpredictable.

[Radonirinaunimi]

Now, everything passed ...

[scarlehoff]

You just need the package to compile in the CI once. Maybe leave an issue open in the feedstock to remind yourself of the issue, but once the package is compiled (and uploaded to conda-forge) the unpredictability doesn't matter anymore.

[Radonirinaunimi]

So, @cschwan, @scarlehoff I guess we can then merge conda-forge/pineappl-feedstock#44?

[cschwan]

I'll merge it.

[cschwan]

I had a closer look into build logs and it seems that the version isn't properly set and therefore curl downloads from the wrong link.

[cschwan]

Very strange, it doesn't seem to be entirely reproducible. Should the error 'tar: Error opening archive: Unrecognized archive format' appear again, check the previous output which looks similar to the folllowing:

prefix:  '$PREFIX'
target:  'x86_64-apple-darwin'
version: '0.8.2'
URL:     'https://github.com/NNPDF/pineappl/releases/download/v0.8.2/pineappl_capi-x86_64-apple-darwin.tar.gz'

Most likely the version field is wrong and therefore the URL also, which leads to the error message.

[cschwan]

@Radonirinaunimi @scarlehoff please check that v0.8.2 works via Conda and then please close the Issue.

[Radonirinaunimi]

That error is indeed very strange. I am not really sure what is going on here.

But the v0.8.2 via conda now works fine, at least all the tests on nnpdf are now passing NNPDF/nnpdf#2129. So, I think we can close the issue.