[wg-camlp4] Editor support for syntax extensions
Török Edwin
edwin+ml-ocaml at etorok.net
Wed Jan 30 11:19:30 GMT 2013
On 01/30/2013 11:12 AM, Alain Frisch wrote:
> On 01/30/2013 03:34 AM, Hongbo Zhang wrote:
>> let rec token enc = {:lex|
>> "<utf8>" -> begin enc := Ulexing.Utf8; token enc lexbuf end
>> | "<latin1>" -> begin enc := Ulexing.Latin1; token enc lexbuf end
>> | xml_letter+ -> Printf.sprintf "word(%s)" (Ulexing.utf8_lexeme lexbuf)
>> | number -> "number"
>> | eof -> exit 0
>> | [1234-1246] -> "bla"
>> | "(" -> begin
>> Ulexing.rollback lexbuf; (* Puts the lexeme back into the buffer *)
>> {| "(" [^ '(']* ")" -> Ulexing.utf8_lexeme lexbuf |} lexbuf
>> (* Note the use of an inline lexer *)
>> end
>> | "(*" -> begin comment lexbuf; "comment" end
>> | ' ' -> "whitespace"
>> | _ -> "???" |}
>> and comment = {:lex|
>> "*)" -> ()
>> | eof -> failwith "comment"
>> | _ -> let _lexeme = Ulexing.lexeme lexbuf in
>> comment lexbuf |}
>
> This looks very bad to me:
>
> 1. You loose all support from your editor (indentation, coloring, parentheses matching). Am I the only one who finds this really problematic? With quotations, my emacs looks like notepad...
I still get syntax highlighting and matching parens in both vim and emacs for the above (although maybe antiquotations would confuse it):
http://www.pasteall.org/pic/show.php?id=44586
http://www.pasteall.org/pic/show.php?id=44587
I do agree that editor support is less reliable once you start using syntax extensions. A common way to define quotations and anti-quotations could probably
help editors cope with syntax extensions better though.
> Indentation and coloring do find typos on the fly; automatic indentation makes it easy to copy/paste code from one context to another one.
Editor support is something that I meant discuss.
1. Auto-completion support.
This already exists for Emacs in the form of Typerex's ocp-complete I think.
It might be possible to adapt it for Vim, for example see how ClangComplete works here for C: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/vim/vimrc?view=markup.
I'd like to have this idea extended to syntax extensions too, where the syntax extension *itself* provides the possible completions.
For example: a lexer/parser extensions could provide completion based on available tokens, a monadic syntax extensions would provide completion
based on available monad variables, etc.
And of course once you go back to original OCaml syntax inside a quotation it should provide the regular OCaml completions.
2. Syntax highlighting support
Currently each editor has its own syntax highlighting rules, but there are some (corner) cases
where even with no syntax extensions it gets it wrong (ok, maybe I should file bugs about those).
What I had in mind is that (optionally) the Vim syntax highlighter script passes the source code through an external process and gets back
markers for syntax highlighting, and it uses that to override its built-in highlighter:
@1,5-1,17 Keyword
@1,17-3,28 Comment
...
Not sure how hard would it be to write such a Vim (or Emacs) script, and how well it'd work in practice (you probably
wouldn't want to do this for every character the user types, maybe only do this on file load/during idle times/after entering a complete line/saving/etc.).
This is again probably similar to what Typerex already does for the OCaml syntax.
As for the external process, it'd use compiler-libs to parse the source code, run any -ppx/camlp4 extensions on it, take the locations from the original parsetree
and output these syntax markers.
To take your IFDEF example:
| {pmod_desc = Pmod_apply(
{pmod_desc = Pmod_apply(
{pmod_desc = Pmod_apply(
{pmod_desc = Pmod_ident {txt = Lident "IFDEF"}, pmod_loc = ifdef_loc},
{pmod_desc = Pmod_ident {txt = Lident sym}}
)},
body_def)}, body_not_def)} ->
self#highlight ifdef_loc `Keyword;
if getenv sym <> "" then this # module_expr body_def
else this # module_expr body_not_def
....
Then astmapper could have two kinds of outputs:
* by default it outputs the new parsetree
* if there is a '--highlight' command-line it throws away the parsetree and prints instead the highlight marker collected by self#highlight, and then uses default
OCaml highlighting to output markers for parsetree nodes that didn't get self#highlight called.
3. Syntax checking support
Something similar to gcc's -fsyntax-only, but better. Currently you need to compile a fake .cmi or ask for a fake .inferred.mli to get syntax + type errors for the current file,
which is time consuming sometimes.
The above tool could have a flag to also output syntax errors (since it has to parse the source code anyway, and it knows what syntax extensions are used), _and_
it could also try to show type errors based on the current file, and any already existing .cmi files.
4. What syntax extensions to use => define in the source code
This has already been mentioned that the source file could define what syntax extensions it uses, so it works for all build systems (ocamlbuild, makefiles, omake), etc.
It may not even need special support from the compiler: there could be a -ppx that finds these pragmas and loads those other -ppx on its own!
Best regards,
--Edwin
More information about the wg-camlp4
mailing list