The `helm-unittest` updatecli target should be of type `file` instead of `yaml`

[lemeurherve]

The target

helm-charts/updatecli/updatecli.d/helm-unittest.yaml

Lines 25 to 32 in 3e42c70

	targets:
	updateChart:
	name: Update helm-unittest version in the GitHub Action workflow
	kind: yaml
	scmid: default
	spec:
	file: .github/workflows/test.yml
	key: $.env.UNITTEST_VERSION

should be of type file to avoid updatecli YAML parser adding a null after workflow_dispatch while it shouldn't: #706 (comment)

[dduportal]

Is that a problem?

I see that the workflow_dispatch is still enabled:

It looks like it is valid. Using file means having a regexp which is a PITA to maintain, so better to keep updatecli doing this if it is valid YAML + GHA syntax no?

[lemeurherve]

I've removed the null manually. For GitHub workflow_dispatch: is valid and the correct syntax, while the YAML parser used by updatecli wants to add a null to it: workflow_dispatch: null

[dduportal]

• There are no differences in strict YAML between workflow_dispatch:, workflow_dispatch: null or even ``workflow_dispatch: Null: that is why the parser is correct, but it's *visually* annoying. Ref. the YAML specification for null` https://yaml.org/type/null.html

• There are no differences in GHA (functionally speaking) between workflow_dispatch:, workflow_dispatch: null and workflow_dispatch: {}: the 3 syntaxes enable the button to trigger manually a build. So wether you've removed the null value or not , the feature is correct: hence my question "what is the problem you want to solve"

Again, using a file with a pattern is hard to keep (the amount of fixes we've had to fix over time...) compared to a native YAML system. If there are no problem, I don't understand why we should trade the native resource with another?

[hervelemeur]

As you said, it's visually annoying, that's all.

As there is only one dependency to keep up to date, the compromise didn't seem too big for me.

Your explanations are clear, closing this issue.