[Experiment] - Look into compiling feelers to feel directly to enable more broader uses of the templating language

[Skaiir]

What should we do?

We should investigate into compiling feelers to feel, this could allow us to use feelers any place where feel is currently relied upon.

Why should we do it?

The value of this would be to plug in feelers into any system that uses feel by removing the need for feelers proper to be evaluated, for example in the backend of some of our projects.

[nikku]

A prototype of how this could look like has been shared here:

@Skaiir and myself discussed this topic briefly, the merits of "transpiling to a known language" and the impacts on editor tooling:

• Similar to what we learned "from the web", the moment you transpile you
  • (1) critically rely on editor tooling
  • (2) risk a split brain situation
  • (3) will have a problem with debugging
• You immediately start to discuss "source maps" as a result

---

I opened camunda/camunda-modeler#5202 to record an alternative - proper string templating support, across our stack.