Support BER MetaOCaml N111 (in Unix)

[moatom]

This PR is for supporting BER MetaOCaml N111, which is OCaml 4.11.1 extended to support meta-programming (see http://okmij.org/ftp/ML/MetaOCaml.html). Because MetaOCaml adds >., .~, and .< to original OCaml syntax, it adds GREATERDOT, DOTTILDE, and DOTLESS to token.

I tested it but tools/test_branch.sh ber-n111 did't work, of course because HEAD was not buildable in MetaOCaml.
The test by make test worked fine :-)

[hhugo]

I'm surprised this is enough. Don't you want to print theses new tokens back ?

[moatom]

Hi!

Don't you want to print theses new tokens back ?

Are you mentioning about OCamlFormat's changing quoting syntax sugars to attributes representations (e.g. .<1>. -> 1[@metaocaml.bracket]) ? Yes, I do. But I don't know how to do it.

[moatom]

Although I am not familiar with the entire structure of OCamlFormat, I think Printast.expression's difference is related to this problem.

• https://github.com/ocaml-ppx/ocamlformat/blob/master/lib/Migrate_ast.mli#L124
• https://github.com/metaocaml/ber-metaocaml/blob/ber-n111/parsing/pprintast.ml#L639
• https://github.com/ocaml-ppx/ppxlib/blob/master/ast/pprintast.ml#L597

[hhugo]

A special case will need to be added early in https://github.com/ocaml-ppx/ocamlformat/blob/master/lib/Fmt_ast.ml#L1553.
similar to what is done in metaocaml.
Note that Metaocaml uses attributes to encode theses new constructs. It is not ideal as it makes attributes order dependent, see discussion in ocaml/ocaml#10131

[craigfe]

You may be interested in the corresponding issue for BER MetaOCaml support in OCamlFormat, and this patch that provides a minimal implementation with the current use of attributes.

[hhugo]

You may be interested in the corresponding issue for BER MetaOCaml support in OCamlFormat, and this patch that provides a minimal implementation with the current use of attributes.

You may be interested in the corresponding issue for BER MetaOCaml support in OCamlFormat, and this patch that provides a minimal implementation with the current use of attributes.

Nice
Note that your patch doesn't not care about the order of attributes. I think it will make both .< .~x >. and .~(.<x>.) format the same

[craigfe]

Indeed, thanks.

For what it's worth, I agree with your comments on the corresponding OCaml PR that extension points would be a better representation here & think it's worth waiting a little bit on the OCamlformat side until that question is resolved upstream.

[moatom]

You may be interested in the corresponding issue for BER MetaOCaml support in OCamlFormat, and this patch that provides a minimal implementation with the current use of attributes.

Thanks! It was very easy to understand.

think it's worth waiting a little bit on the OCamlformat side until that question is resolved upstream.

I roughly understand the current situation. I think so, personally.

[hhugo]

metaocaml support added in #2630