Conversation
|
|
||
| const artifactClient = new DefaultArtifactClient(); | ||
| const artifactName = hex; | ||
| const artifactName = isEmptyInput(artifact_name) ? hex : artifact_name + matrix_key; |
There was a problem hiding this comment.
I hesitated here to rename the artifactName const to something else to differentiate them. However since I noticed inputs are in snake case, I thought maybe that's your convention to differentiate inputs form the rest and that's enough
|
Ping. Is there any progress on this PR? It not only fixes #44 but also allows to support workflow re-runs. Without the patch, change in content of the file will result in a different file name. This means that in case the job is re-run (because it failed in the first attempt) there will be 2 result files stored in the workflow. When reading the results random file will be selected. Since there are 2 files with the same key and different values for this key, user can't trust in the output of the |
2 participants
what
An optional input
artifact-nameis added to allow the user to specify how to name the artifacts stored.If the input is missing, the hash is used as before.
why
Due to the use of a hash of the content as an artifact name, if different steps (with a different step name) try to use the action to write the same content, the build will fail. The first step will succeed but the second one will fail because an artifact with the same name already exists.
references
Fixes #44