2025/02/25

Newest at the top

2025-02-26 00:38:43 +0100merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) (Ping timeout: 245 seconds)
2025-02-26 00:37:49 +0100rvalue(~rvalue@user/rvalue) rvalue
2025-02-26 00:36:46 +0100__monty__(~toonn@user/toonn) (Quit: leaving)
2025-02-26 00:32:08 +0100merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) merijn
2025-02-26 00:28:14 +0100alfiee(~alfiee@user/alfiee) (Ping timeout: 252 seconds)
2025-02-26 00:23:46 +0100alfiee(~alfiee@user/alfiee) alfiee
2025-02-26 00:21:03 +0100merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) (Ping timeout: 244 seconds)
2025-02-26 00:20:18 +0100rvalue(~rvalue@user/rvalue) (Quit: ZNC - https://znc.in)
2025-02-26 00:16:46 +0100merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) merijn
2025-02-26 00:12:53 +0100 <tomsmeding> but yes, a splitting function is perhaps nicer here
2025-02-26 00:12:43 +0100 <monochrom> Although, I confess that I want 0 dependencies so I wrote my own recursion :)
2025-02-26 00:12:40 +0100 <tomsmeding> hololeap: https://paste.tomsmeding.com/QHaiHEVF
2025-02-26 00:12:12 +0100 <monochrom> You can write your own recursion over span. But the split package does that for you.
2025-02-26 00:11:39 +0100 <monochrom> Oh, the split package has tools for that too.
2025-02-26 00:11:37 +0100 <tomsmeding> if you write out that <$> <*> sequence as a do-block, it gets much more readable
2025-02-26 00:10:57 +0100 <hololeap> although I wonder if it would be more readable to just do this with span from Data.List
2025-02-26 00:10:44 +0100 <monochrom> BTW the "convenience" of regex allowing that is avenged by the fact that such high non-determinism causes exp-time or exp-space.
2025-02-26 00:10:43 +0100 <tomsmeding> monochrom: thanks, it's been too long since I parsec'd, this is also nicer to write
2025-02-26 00:10:32 +0100 <hololeap> cool, thanks :)
2025-02-26 00:10:29 +0100foul_owl(~kerry@174-21-138-88.tukw.qwest.net) foul_owl
2025-02-26 00:10:20 +0100 <tomsmeding> hololeap: there
2025-02-26 00:10:09 +0100 <yahb2> Right [("this","that"),("black","white")]
2025-02-26 00:10:09 +0100 <tomsmeding> % parse (sepBy ((,) <$> (many (satisfy (/= ':')) <* char ':') <*> many (satisfy (`notElem` ":,"))) (char ',')) "<stdin>" "this:that,black:white"
2025-02-26 00:09:49 +0100 <monochrom> What you should really lament, if you like, is that parsec is not regex, so whereas in regex you just have to say ".*,.*", you simply can't do that in parsec.
2025-02-26 00:08:59 +0100 <tomsmeding> I guess
2025-02-26 00:08:50 +0100 <monochrom> many (satisfy (/= ',')) pretty please
2025-02-26 00:08:26 +0100 <monochrom> Yikes, abuse of lookAhead again.
2025-02-26 00:08:23 +0100 <yahb2> <interactive>:289:99: error: [GHC-83865] ; • Couldn't match type ‘()’ with ‘Char’ ; Expected: ParsecT ; String () GHC.Internal.Data.Functor.Identity.Identity Char ; ...
2025-02-26 00:08:23 +0100 <tomsmeding> % parse (sepBy ((,) <$> manyTill anyChar (char ':') <*> manyTill anyChar (lookAhead (oneOf ":," <|> eof))) (char ',')) "<stdin>" "this:that,black:white"
2025-02-26 00:08:07 +0100 <tomsmeding> evidently not!
2025-02-26 00:07:56 +0100 <yahb2> <interactive>:287:23: error: [GHC-83865] ; • Couldn't match expected type: ParsecT ; String () GHC.Internal.Data.Functor.Identity.Identity a1 ; ...
2025-02-26 00:07:56 +0100 <tomsmeding> % parse (sepBy ((,) <$> manyTill (char ':') <*> manyTill (lookAhead (oneOf ":," <|> eof))) (char ',')) "<stdin>" "this:that,black:white"
2025-02-26 00:07:41 +0100 <tomsmeding> hm, let's see if I can whiteboard code
2025-02-26 00:07:24 +0100 <yahb2> Right ["abc,def,ghi"]
2025-02-26 00:07:24 +0100 <tomsmeding> % parse (sepBy (many anyChar) (char ',')) "<stdin>" "abc,def,ghi"
2025-02-26 00:07:16 +0100 <yahb2> <no output>
2025-02-26 00:07:16 +0100 <tomsmeding> % import Text.Parsec
2025-02-26 00:06:24 +0100 <tomsmeding> hololeap: also, that "sepBy (many anyChar) (char ',')" wouldn't even work: the `many anyChar` would consume the entire input from here, and because it consumed input, it would count as "succeeded" which means it doesn't backtrack
2025-02-26 00:05:52 +0100merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) (Ping timeout: 252 seconds)
2025-02-26 00:05:03 +0100 <tomsmeding> of course you technically can just start a new parser for that string, and you can also `setInput`, but typically both are a bad idea
2025-02-26 00:04:22 +0100 <hololeap> I see
2025-02-26 00:03:56 +0100 <tomsmeding> you can't "re-inject" some arbitrary string into the source to be parsed
2025-02-26 00:03:13 +0100ljdarj1ljdarj
2025-02-26 00:03:13 +0100ljdarj(~Thunderbi@user/ljdarj) (Ping timeout: 248 seconds)
2025-02-26 00:03:04 +0100 <tomsmeding> also, while you can process the result of parsing later using (>>=), parsec always parses _from_ the input string
2025-02-26 00:02:55 +0100 <hololeap> right, s/anyChar/(many anyChar)/
2025-02-26 00:02:28 +0100 <tomsmeding> anyChar parses _one_ character
2025-02-26 00:02:18 +0100 <hololeap> hm, you would think with Parsec being a monad, it could be done with: sepBy anyChar (char ',') >>= mapM (...)
2025-02-26 00:01:26 +0100ljdarj1(~Thunderbi@user/ljdarj) ljdarj
2025-02-26 00:01:24 +0100merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) merijn