fix(task): task run finished not consistent#304
Conversation
This commit task run to conform to other outcome values being enums with values: "success", "failure", "error", and "cancel" Signed-off-by: xibz <impactbchang@gmail.com>
| "outcome": { | ||
| "type": "string" | ||
| "type": "string", | ||
| "enum": ["success", "failure", "cancel", "error"] |
There was a problem hiding this comment.
Tests typically have assertions, which may succeed or fail, and in case the tests fails to execute before the assertion is evaluated, that's considered an error. How does that model apply to task runs?
There was a problem hiding this comment.
The same as how it applies to pipeline run. If pipeline run has it as enum, this should probably be an enum was my thinking
There was a problem hiding this comment.
Oh, I see, it was added in #238.... well, I'm ok to make things consistent, but only if there's a meaning - I guess I already expressed by concerns at the time in #237.
I'm fine to keep the two type of failures for TaskRuns and PipelineRuns if there is a meaning to it.
For Tekton I wouldn't know which one to use. I guess we could use one for validation errors (so it didn't run at all) and another one for runtime errors, although the distinction is arbitrary as validation error, when using dynamic values in matrices and results, can also happen at runtime.
As longs as there's clear guidance for the meaning I'm ok. If not I would rather remove it from PipelineRuns.
There was a problem hiding this comment.
Yes, and that is a very important concern. The idea (my plan) is to add these to the _defs so it is very clear on what these mean. But holding off, until we define the subjects that encompass this. Ill create an issue for that, if that helps
2 participants
This commit task run to conform to other outcome values being enums with values: "success", "failure", "error", and "cancel"