As long eslint is using espree, it won't be easy (requires some tweaking).

The parser is configurable: http://eslint.org/docs/configuring/#specifying-parser

If that's not enough, let me know what you need.

@nzakas If that is enough to run ESLint unit tests, I think it should be sufficient.

It probably depends on how you run the tests, but I think it should work. We set the parser in this file: https://github.com/eslint/eslint/blob/master/conf/eslint.json

@nzakas only problem is that the suite currently relies on behavior only espree provides.

If we're dedicated to doing this, it would make sense for us to begin work on converging the two projects, at least for the parts that aren't add ons like jsx.

Outside of JSX and globalReturn, the rest are ES6 features that Esprima harmony already supports.

Right, my point being we'd need a way to ignore those tests. Thoughts?

@mikesherov It might need a simple pre-processing step, e.g. removing files for unsupported tests.

Can we add jstransform and 6to5 in that list too?
(I suppose we need to just get to writing code and pull requesting here...)

Exactly. @jeffmo add at to the description at will, and let's turn it into a checklist :-)

recast, although it is using esprima-fb by default (don't know if that's an issue?)

@fkling how can we configure it to use regular esprima?

@mikesherov: its parse method accepts an esprima option. However, the tests don't seem to be configurable, but I guess one could simply alias esprima-fb to esprima in the test environment. cc @benjamn who wrote recast.

Happy to make the tests configurable, but not immediately sure what the best way to do that is…

Also, if jstransformand recast have some tests requiring features not yet supported, then we need a mechanism to exclude those tests.

Oh right, I forgot that these CI tests will need to run with master -- maybe we should hold off on adding them to CI until master is close enough to harmony that we can switch esprima-fb over to tracking master (and then perhaps it would make more sense to just make esprima-fb include recast and jstransform in it's forked form of the CI tests)

perhaps it would make more sense to just make esprima-fb include recast and jstransform in it's forked form of the CI tests

That's what I was thinking as well.