github actions

[matbesancon]

adding github action for CI and code coverage

[matbesancon]

ping @joaquimg

[matbesancon]

I think CI is fairly fast on GH-actions. Worst thing we can remove them later

[matbesancon]

Since MOI has binary dependencies, then all depending packages also have some, or do you mean direct dependencies?

On Mon, Dec 7, 2020, 19:42 Oscar Dowson notifications@github.com wrote:

@odow commented on this pull request.

In .github/workflows/ci.yml https://github.com/jump-dev/MatrixOptInterface.jl/pull/11#discussion_r537739944 :

• arch: x64
• • version: '1'
• os: ubuntu-latest
• arch: x64
• • version: '1.0'
• os: ubuntu-latest
• arch: x64
• • version: '1.0'
• os: ubuntu-latest
• arch: x86
• • version: '1'
• os: windows-latest
• arch: x64
• • version: '1'
• os: macOS-latest
• arch: x64

The rule-of-thumb I went with was:

• Does the package have binary dependencies?
  • If yes, test on Linux, Mac, and Window 64-bit, Julia 1 and Julia 1.0
    • Also test on linux 32-bit Julia 1.0 for Int32 mistakes

    • https://github.com/jump-dev/GLPK.jl/blob/master/.github/workflows/ci.yml

  • If no, test on Linux 64-bit, Julia 1 and 1.0, and Linux 32-bit, Julia 1.0
    • Optionally test on Window 64-bit, Julia 1.0 to test windows installation

    • https://github.com/jump-dev/MathOptInterface.jl/blob/master/.github/workflows/ci.yml

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jump-dev/MatrixOptInterface.jl/pull/11#pullrequestreview-546431027, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB2FDMXNSX6HIWG2VJ2UOFLSTUO2RANCNFSM4UQM3X4A .

[odow]

Usually just direct. You can assume the down-stream dependencies are well-tested, so you just want to test the first-order integration with binaries (e.g., solvers).

[matbesancon]

ping @joaquimg for merging this one :)

[joaquimg]

Nice!