Experimenting with passing matrix outputs from previous matrix jobs #4

…s set of matrix jobs I need a way to reuse the tags, image names, contexts which are parameters used for building the image. Currently these are available in the matrix jobs but I need to access these in the subsequent jobs so that I can rebuild images from cache. Using custom github action for this: Write outputs: https://github.com/cloudposse/github-action-matrix-outputs-write Read outputs: https://github.com/cloudposse/github-action-matrix-outputs-read

Got an error but values are being passed between steps of matrix jobs. Perhaps missing double quotes was the problem, let's see if that resolves it. ------ Got this error: Error: Outputs should be valid YAML --------------------- host_namespace: ghcr.io/everest/everest-demo image: mqtt-server context: ./mosquitto tags: ghcr.io/everest/everest-demo/mqtt-server:0.0.16 ghcr.io/everest/everest-demo/mqtt-server:latest labels: org.opencontainers.image.created=2024-07-01T17:35:09.789Z --------------------- YAMLParseError: Implicit keys need to be on a single line at line 5, column 1: tags: ghcr.io/everest/everest-demo/mqtt-server:0.0.16 ghcr.io/everest/everest-demo/mqtt-server:latest ^ ----------

So the quotes weren't the issue actually but the fact that the tags passed from the meta step were multiline values. Hence using block level scalar >- to let YAML interpret them as multi-line commands.

Still got the same error for multiline values when using >- YAMLParseError: Implicit keys need to be on a single line at line 6, column 1: ghcr.io/everest/everest-demo/mqtt-server:0.0.16 ghcr.io/everest/everest-demo/mqtt-server:latest ^

The direct access to tags and labels was giving errors since tags and labels were multiline values whereas the matrix output custom action needed single line values for the YAML parser to correctly parse the parameters. I found that we can access the tags and labels via the json object which is an output from the meta action. So, trying that approach now. ------- Another possible approach that I came across, but didn't try yet. I found that with the meta action, we can specify the delimiter / separator type for tags and labels: https://github.com/docker/metadata-action#inputs While tags can have a CSV datatype in docker build push action, labels can only have a List datatype: https://github.com/docker/build-push-action?tab=readme-ov-file#inputs Hence I believe the aforementioned approach would not work. -------

The custom write and read seems to have passed. But error when trying to read data in the env variables part. Error: The template is not valid. .github/workflows/cicd.yaml (Line: 138, Col: 26): A mapping was not expected ----- Now trying to output results directly first, let's see.

…e / read action JSON object data is simply passed as "Object" for tags_labels. Expected list of tags and labels.

…rest-demo into automate-tests-2

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Experimenting with passing matrix outputs from previous matrix jobs #4

Experimenting with passing matrix outputs from previous matrix jobs #4

Commits on Jul 1, 2024