MicrosoftDocs/vsts-rest-api-specs

schema validation fails to on checkout, download and publish steps, as well as parallel count

Opened this issue · 3 comments

david-vankampen commented

If you use the schema returned by this API call to check a yaml file generated by the previewRun REST API call here, you will get failures on a few key steps if they are part of your pipeline:

  1. checkout (shows up as 6D15AF64-176C-496D-B583-FD2AE21D4DF4@1)
  2. download (shows up as 30f35852-3f7e-4c0c-9a88-e127b4f97211@1)
  3. publish (shows up as ECDC45F6-832D-4AD9-B52B-EE49E94659BE@1)
  4. pool->strategy->parallel (expects a string, but previewRun returns an integer)

Is there a recommended transformation we should apply to the output of previewRun to get it to pass schema validation?

I'm not sure if this is more a bug with previewRun or with yamlSchema but I figured I'd start here.

Oddly enough, it seems to replace the script tasks with cmdLine@V2, which helps those not fail schema checking. If previewRun could do that for the 3 items above, that would help a lot!

Document Details

⚠ Do not edit this section. It is required for learn.microsoft.com ➟ GitHub issue linking.

david-vankampen commented

Here is another MSFT document that details which task GUIDs apply to which task. Notably, the guid for download does NOT match what previewRun is currently providing (though neither would pass the schema validation).

david-vankampen commented

Here is a pretty minimal version of a pipeline yaml file that can create the issues mentioned:

pool:
  vmImage: ubuntu-latest
strategy:
  parallel: "2"


steps:
- checkout: self
  submodules: recursive
- publish: azure-pipelines.yml
  artifact: exampleartifact
- download: current
  artifact: exampleartifact

when you call previewRun, the exports as

{
    "stages": [
        {
            "stage": "__default",
            "jobs": [
                {
                    "job": "Job",
                    "pool": {
                        "vmImage": "ubuntu-latest"
                    },
                    "strategy": {
                        "parallel": "2"
                    },
                    "steps": [
                        {
                            "task": "6d15af64-176c-496d-b583-fd2ae21d4df4@1",
                            "inputs": {
                                "repository": "self",
                                "submodules": "recursive"
                            }
                        },
                        {
                            "task": "ecdc45f6-832d-4ad9-b52b-ee49e94659be@1",
                            "inputs": {
                                "path": "azure-pipelines.yml",
                                "artifactName": "exampleartifact"
                            }
                        },
                        {
                            "task": "30f35852-3f7e-4c0c-9a88-e127b4f97211@1",
                            "inputs": {
                                "alias": "current",
                                "artifact": "exampleartifact"
                            }
                        }
                    ]
                }
            ]
        }
    ]
}

which will fail schema compliance checking, because none of those tasks exist in the schema.

david-vankampen commented

In order to pass schema compliance checking, I had to turn the above steps into this:

"steps": [
    {
        "checkout": "self",
        "submodules": "recursive"
    },
    {
        "task": "PublishPipelineArtifact@1",
        "inputs": {
            "targetPath": "azure-pipeline.yml"
        }
    },
    {
        "task": "DownloadPipelineArtifact@1",
        "inputs": {
            "artifactName": "exampleartifact"
        }
    }
]

which is a non-trivial transformation. You see checkout lost the inputs object, for publish and download, it had to drop the alias input property, and for publish artifactName had to become targetPath , and for download, artifact had to become artifactName. I would think it should export this way, not requiring an after-the-fact transform to make it schema-compliant.