Mbedtls

[izahn]

Checklist

• Used a personal fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

Closes #125

Note that this needs conda-forge/staged-recipes#16113 , once someone reviews and merges that we can proceed to review this PR.

[conda-forge-linter]

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.

[mkitti]

mbedtls does not exist in conda-forge.

[mkitti]

Why did you remove the pin?

[izahn]

I used https://github.com/archlinux/svntogit-community/blob/packages/julia/trunk/julia-hardcoded-libs.patch to patch out the hard-coded mbedtls version. Since that patch also removes suitesparse version hard-coding we don't need to pin anymore.

[mkitti]

I would be interested to see how this interacts with the issue in #118, which is why the pin was added.

[izahn]

Yes, we should test to make sure that issue is fixed by this PR before merging.

[mkitti]

Which package are you referring to? mbedtls does not exist in conda-forge

[izahn]

mbedtls does not exist in conda-forge.

As noted in my PR it will once conda-forge/staged-recipes#16113 is merged.

[izahn]

@conda-forge-admin please restart ci

[mkitti]

For Mac, we need to resolve the problem described in this comment when using system LibGit2:
#115 (comment)

[izahn]

For Mac, we need to resolve the problem described in this comment when using system LibGit2:
#115 (comment)

That thread is pretty long, can you summarize what the problem is?

[mkitti]

That thread is pretty long, can you summarize what the problem is?

I linked to a specific comment in the thread. Here's the gist.

Changing from USE_SYSTEM_LIBGIT2=0 to USE_SYSTEM_LIBGIT2=1 we introduce an error on Mac OS X.

Error During Test at $PREFIX/share/julia/stdlib/v1.6/Downloads/test/runtests.jl:12
  Got exception outside of a @test
  error setting certificate verify locations:  CAfile: /Users/runner/miniforge3/conda-bld/julia_1628019948829/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placeho while requesting https://httpbingo.julialang.org/base64/SnVsaWEgaXMgZ3JlYXQh
  Stacktrace:
    [1] (::Downloads.var"#9#18"{IOStream, Base.DevNull, Nothing, Vector{Pair{String, String}}, Float64, Nothing, Bool, Bool, String, Int64, Bool, Bool})(easy::Downloads.Curl.Easy)
      @ Downloads ~/miniforge3/conda-bld/julia_1628019948829/work/usr/share/julia/stdlib/v1.6/Downloads/src/Downloads.jl:356

[mkitti]

Essentially we need USE_SYSTEM_LIBGIT2=1 for all platforms. It currently does not work on OS X because of "error setting certificate verify locations". I'm not sure what that error is about.

[izahn]

Essentially we need USE_SYSTEM_LIBGIT2=1 for all platforms. It currently does not work on OS X because of "error setting certificate verify locations". I'm not sure what that error is about.

I haven't had any luck solving this one. Does using libgit2 from BinaryBuilder on osx cause any known problems?

[izahn]

I cleaned this up and force-pushed, should be in good shape now. I also added explicit tests to make sure that this PR successfully addresses both #125 and #118

[izahn]

This PR is ready for review. It fixes #125 and #118 by applying a patch adapted from the Arch Linux Julia package and using the system mbedtls on linux.

[izahn]

@conda-forge-admin please rerender

[izahn]

@conda-forge-admin please rerender

[mkitti]

I also added explicit tests to make sure that this PR successfully addresses both #125 and #118

Where are these explicit tests?

[izahn]

I also added explicit tests to make sure that this PR successfully addresses both #125 and #118

Where are these explicit tests?

Thanks for catching that! Somehow they didn't make it in. I pushed them up in 77a8d02

[mkitti]

Things are looking great. I'm just trying to reduce the changes to the minimum. Currently, I think there are some changes that were reverted from #123. I have unreverted them here:
izahn#1

[izahn]

The changes in izahn#1 are all to files generated by conda smithy rerender. What is issue you are trying to address there?

[mkitti]

I'm trying to minimize the actual diff against master.

[mkitti]

I'm ready to merge this after it either does not modify .ci_support versus master or we can justify why this pull request should modify .ci_support versus master.

[izahn]

I'll back it up and let the bot do the rerender so we can see what happens.

[izahn]

5b15552 just changes the recipe, none of the CI or migration stuff. Now we rerender:

@conda-forge-admin please rerender

[izahn]

The bot ran conda smithy rerender for us in 3820715 . All perfectly normal and expected. Rerendering is standard part of the conda packaging workflow, changes produced by it do not need to be reverted.

[github-actions]

Hi! This is the friendly conda-forge automerge bot!

I considered the following status checks when analyzing this PR:

• linter: passed
• azure: passed

Thus the PR was passing and merged! Have a great day!

[ngam]

Essentially we need USE_SYSTEM_LIBGIT2=1 for all platforms. It currently does not work on OS X because of "error setting certificate verify locations". I'm not sure what that error is about.

I haven't had any luck solving this one. Does using libgit2 from BinaryBuilder on osx cause any known problems?

@izahn @mkitti we can try to see if the new libgit2 patch can resolve this for macs. I will start a new PR. Follow #155