add template engine functionality to support generic templates

[galargh]

Template engine

I created a README.md that describes the template engine from the end user perspective i.e. what the syntax is, what objects are available, etc.

On the backend, I use perl to:

1. perform a regex match on \${{{(.?*)}}};
2. execute jq with the matched group as a filter and context JSON object;
3. replace the match with jq output

Extra files

The newly added extra_files config field can be used to request copying of extra files while not modifying the files value from defaults.

Use-case example

See #434

Testing

• deployed new workflow to .github-test-target from testing: dispatch + copy-workflow + PR

[marten-seemann]

Is the sync-release-assets the only use case for this?
It seems like this could be easier accomplished by just publishing a GH Actions Workflow and manually calling this workflow from the two (?) repos that care about it, no?

[marten-seemann]

https://xkcd.com/1171/

[galargh]

Is the sync-release-assets the only use case for this?

It is a concrete example that I decided to use when working on templating functionality but it is not the only use case. Deploying issue templates is another example that was mentioned here #57

It seems like this could be easier accomplished by just publishing a GH Actions Workflow and manually calling this workflow from the two (?) repos that care about it, no?

Yes, it is possible to manually copy the workflow file to interested repositories to achieve the same but I believe it is a pretty good fit for this repository to be the one distributing it. At the moment there are only 2 but I think there might be more to come. @lidel might be able to elaborate on this.

[laurentsenta]

2 notes:

• I couldn't get an exit code != 0 if the templated variable was missing, could be my env when I tested, the expansion was "just" null without error.
• Do we have a complexity threshold when we want to move these to "something other than shell scripts"?

[galargh]

• I couldn't get an exit code != 0 if the templated variable was missing, could be my env when I tested, the expansion was "just" null without error.

That's expected and that's fine for what it's trying to be right now.

• Do we have a complexity threshold when we want to move these to "something other than shell scripts"?

I think if we ever need to develop the functionality further, we should consider alternatives.

I've been able to simplify the setup by moving the config to env section of the step. Tested: https://github.com/protocol/.github/actions/runs/4064327038