Reason distillery

[charlesetc]

This is building off of #445 by @vasilisp.

It's adding the -reason flag to eliom-distillery which converts the .eliom files to reason syntax. It also adds the %%%REASON_FLAG%%% substitution which evaluates to "-reason" if -reason is passed to eliom-distillery and an empty string otherwise.

Hopefully this gets ocsigen closer to reason support - I haven't been able to get it working with ocsigen-start, but this does generate a reason ocsigen program with:

eliom-distillery -name project_name -template basic.ppx -reason

The problems current with getting ocsigen start to work are in the translation from the ocaml to reason.

• There is some camlp4 for postgres. This only affects 2 files. And there are some ppx forks of pgocaml but they aren't on opam
• The syntax ##. doesn't seem to work when converting from ocaml to reason, it's used in js_of_ocaml ppx

It's possible these problems are too great right now - we might want to wait to merge the -reason flag until there's support for ocsigen start. Either way, it's a little closer to reason and ocsigen! ✈️

[Drup]

I'm surprised by the ##. issue. Are reason people aware of it ?

[vasilisp]

Thanks!

I have no comments on the code. Looks good.

I like the idea of using refmt during template instantiation.

I am a bit worried about the template/eliom-distillery interface. If we were to add a template using Ocamlbuild/Jbuilder/..., would we be able to interpret your fixed %%%REASON_FLAG%%%? Could this be customized? Or at least, could we allow individual templates to declare support for Reason (or lack thereof)?

[charlesetc]

Hmm that's a good point about the -reason flag being specific to eliomc, etc. I'm not sure how to customize it ... maybe we could change it to REASON_ELIOMC_FLAG to be more specific, or as you say declaring support for reason in the templates makes sense. What do you think?

[vasilisp]

Sorry for the delay.

The customizability of the flag and the declaration of Reason support could both be implemented via a per-template configuration/specification file. It could be useful for other features (e.g., supersede the existing.eliom* files), and we can discuss it later.

For now, could you work-around the use of Filename.extension via Filename.check_suffix (or otherwise)? Let's not break 4.03 compatibility just yet.

Thanks!