Introduce JSON matrix workflow #3013
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
glbrntt
approved these changes
Dec 3, 2024
Introduce and adopt a new method for defining test matrices, `swift_test_matrix.yml`.⚠️ Any external adopters of the unit tests, Cxx interop and benchmarks workflows are automatically opted in to use the new infrastructure. Motivation: * The current matrix workflow has the limitation that it only supports pre-defined sets of variables which are explored in the test matrix. At the moment this is a pre-defined set of Swift versions on Linux and Windows. * Adding more means hard-coding them at multiple levels of the workflow hierarchy. * Currently skipped Windows matrix jobs show up as successes in the GitHub UI leading to a misleading impression of good coverage. Modifications: Introduce and adopt a new method for defining test matrices, `swift_test_matrix.yml`. The new method is based around the approach of defining the test matrix via a JSON object which may be supplied via an input string from another workflow or from a file on-disk in a repository. Taking this approach means that we have the ability to add new targets to the matrix simply by adding new elements to the JSON array, increasing flexibility and the scope for future growth. The unit tests, Cxx interop and benchmarks workflows are all modified to use the new approach, this opts-in all downstream adopters. This should be transparent to all downstream adopters. In order to unify the Linux and Windows jobs I removed the use of the `container:` GitHub Actions affordance in the Linux jobs which transparently means all steps are executed within the nested container. Instead we must manually call in to the docker container which complicates scripting a little. I tested to see if doing this slowed down the jobs (perhaps GitHub was caching the docker images more intelligently) but it does not. Result: * More flexible test matrix definitions * No more false-passes for disabled Windows targets
glbrntt
approved these changes
Dec 4, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Introduce and adopt a new method for defining test matrices,
swift_test_matrix.yml.Motivation:
Modifications:
Introduce and adopt a new method for defining test matrices,
swift_test_matrix.yml. The new method is based around the approach of defining the test matrix via a JSON object which may be supplied via an input string from another workflow or from a file on-disk in a repository.Taking this approach means that we have the ability to add new targets to the matrix simply by adding new elements to the JSON array, increasing flexibility and the scope for future growth.
The unit tests, Cxx interop and benchmarks workflows are all modified to use the new approach, this opts-in all downstream adopters. This should be transparent to all downstream adopters.
In order to unify the Linux and Windows jobs I removed the use of the
container:GitHub Actions affordance in the Linux jobs which transparently means all steps are executed within the nested container. Instead we must manually call in to the docker container which complicates scripting a little. I tested to see if doing this slowed down the jobs (perhaps GitHub was caching the docker images more intelligently) but it does not.This approach follows the pattern of @FranzBusch 's open PR #2942
Result: