Use GH Actions for CI #262
Conversation
These were accidentally deleted when shuffling macOS setup steps to an `macos-setup/action.yml` and back. These variables make Xcode behavior consistent across different GHA images.
|
Sorry, I forgot to finish the macOS build steps (notarization, etc.) so this isn't really ready for review. The option to convert this PR to a draft isn't showing up for me, so I'll remove the review requests for now. I'll re-request when it's ready. |
| inputs: | ||
| configuration: | ||
| description: The configuration to build, like "Debug" | ||
| required: true | ||
| default: Debug | ||
| artifact_tag: | ||
| description: A string to tag the build artifact, like "Debug" or "MAS" | ||
| required: false | ||
| default: "" |
| AC_USERNAME: ${{ secrets.AC_USERNAME }} | ||
| AC_PASSWORD: ${{ secrets.AC_PASSWORD }} |
There was a problem hiding this comment.
Are the AC_USERNAME and AC_PASSWORD the same for scratch-desktop? I have these set in repository secrets as well, but if they are the same we should probably move them to org secrets.
There was a problem hiding this comment.
Yeah, that would make sense!
| @@ -65,7 +65,7 @@ jobs: | |||
| run: | |||
| shell: bash # even on Windows, unless otherwise specified | |||
| env: | |||
| SCRATCH_SHOULD_SIGN: "NO" # TODO | |||
| SCRATCH_SHOULD_SIGN: "YES" # TODO | |||
| MATCH_STORAGE_MODE: git | |||
There was a problem hiding this comment.
This is also in secrets if you want to use the variable FL_STORAGE_MODE
There was a problem hiding this comment.
Haha, yeah, I had that in a local secret for a while, but it made GHA redact all instances of git in the logs, which was just silly. We could add it as a var instead of a secret, but having the GIT_URL secret as well means changing the storage mode by itself wouldn't be enough to reconfigure it. 🤷
Resolves
Resolves ENG-46
Proposed Changes
Implement GitHub Actions CI for Scratch Link and remove CCI config. The actual build steps are extracted into reusable actions to make the main
ci.ymlmore DRY.Supporting changes:
Note that the calculated version number looks incorrect; I believe that's a limitation of using GitVersion for that. I might switch the build-time version calculation to use the value from the
package.jsonfile, but that'll be a separate PR.