OPTIONS and LANGUAGE completion using ghci #444
Conversation
symbol-at-point => thing-at-point 'symbol Also thing-at-point, then bounds-of-thing-at-point looks it up twice, which is unnecessary.
|
The allowed IIRC, those lists already collected elsewhere at startup by quering |
Also, move option/language-pragma lists out of haskell-yas, because it's nothing to do with yasnippet, and they are needed for completion.
|
I see, I found it: it's in haskell-yas, and you have to type "lang" by itself, not inside a comment, and it will expand to a LANGUAGE pragma inside a proper comment and offer completions. I missed that, I didn't look in the yasnippet menu carefully. I can't find an OPTIONS snippet, I think it's not there. I would still like for ordinary non-snippet completion to work for these too. I changed it to precompute the list of completions properly. |
|
@gracjan I didn't remove the snippet, I added completion but it doesn't replace the snippet. |
|
@gracjan those two facilities (snippet vs. completion-at-point) are not mutually exclusive (but rather complementary -- I use both), I just wanted to hint at that there's already infrastructure that could be shared. |
OPTIONS and LANGUAGE completion using ghci
|
I just committed a change renaming |
Similar to a previous PR, this adds completion for LANGUAGE and OPTIONS ghc pragmas inside comments, using ghci's
:complete repl ":set ...", which understands all language/option flags.It's not spectacularly useful because normally (?) you'd need to type the first three (?) characters yourself before completion kicks in.
I also made a small change to some previous code to avoid some unnecessary string -> symbol -> string conversion.