Configurable language parsing

[nikku]

Proposed Changes

This allows us to configure custom language parsing to be used. This way we can validate (using text validation) expression and unaryTests, but also provide custom context.

Related to camunda/camunda-modeler#3983

Checklist

To ensure you provided everything we need to look at your PR:

• Brief textual description of the changes present
• Visual demo attached
• Steps to try out present, i.e. using the @bpmn-io/sr tool
• Related issue linked via Closes {LINK_TO_ISSUE} or Related to {LINK_TO_ISSUE}

[nikku]

I'd be interested in your feedback @barmac and @marstamm: What representation shall we use for context here? The low-level context required by lezer-feel or the higher level context (variables, built-ins) provided by feel-editor?

[marstamm]

I think that depends on how we want to use it. We use it in bpmnlint-plugin-camunda-compat, where we currently don't have any concept of available Context. Do we want to duplicate the built-ins in linting as well, similar to how we do it in the editor?

[barmac]

Some use cases I can see for this tool in DMN editor are:

• detect usage of unavailable variables
• type mismatch (e.g. variable has type number but function expects string)
• function arguments mismatch (e.g. function expects 2 parameters but 3 were given)

To support this, the context should not assume that we can pass exact values.

[nikku]

Thanks for your feedback folks. What I get is the following:

• We want to have a unified way to provide context, to feel-lint and feel-editor, not based on actual data, but based on the data model.
• We shall consider to also adopt this format for lezer-feel, where it is equally needed. For example in lezer-feel you cannot type functions right now, but what you'd want to express (grammar wise) is function(arg1: type, args: type) : return_type.

[barmac]

I am making it a draft as it's still in progress.

Made on wrong PR