Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-25

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:18 kaare_ joined #marpa
01:34 ronsavage Also re licences. I found this in the Makefile.PL for ExtUtils::MakeMaker: http://dev.perl.org/licenses/.
01:35 ronsavage I'm going to put that URL in the Makefiles of my modules, if appropriate.
04:50 idiosyncrat_ Deep into refactoring Marpa::R3.
04:50 idiosyncrat_ Good night!
05:57 ronsavage joined #marpa
07:13 gleki joined #marpa
08:28 VsyachePuz joined #marpa
11:20 kaare_ joined #marpa
11:55 gleki Hi. So Mapra doesn't have negative lookahead?
12:08 gleki aredridel: hi!
12:13 JPGainsborough joined #marpa
13:12 Pursuit joined #marpa
16:22 idiosyncrat_ joined #marpa
16:23 idiosyncrat_ gleki: negative lookahead == ability to look backward?  Sure does.
16:24 idiosyncrat_ http://search.cpan.org/~jkegl/Ma​rpa-R2-3.000000/pod/Progress.pod
16:25 idiosyncrat_ Marpa is "left-eidetic" meaning that it knows every possible parse so far, which is a feature unique to it.
16:41 gleki idiosyncrat_: well, I mean like "!" in PEG.
16:41 gleki E.g. consider you have a string "letter letter letter" where letter is any terminal but you need to exclude too consecutive letters.
16:41 gleki so "a" "a" would be impossible.
16:44 gleki And I wonder how I can do that.
16:44 idiosyncrat_ Implementing that is a hard problem, because it's not BNF and its not clear what it means.  PEG does it in two ways:
16:45 idiosyncrat_ 1.) By not actually delivering the grammar described: http://jeffreykegler.github.io/Ocean-of-A​wareness-blog/individual/2015/03/peg.html
16:45 idiosyncrat_ 2.) By staying deterministic, and severely limiting the grammars it can handle.
16:45 gleki why not clear? "aa" should prasefail but "ab", "ac", "ad" ... should succeed.
16:45 gleki *parsefail
16:46 gleki so do u have any ideas hwo to do that? In PEG I just use "a !a"
16:47 idiosyncrat_ Yes, the PEG "!" will sometimes work as described in which case you're in luck.
16:47 idiosyncrat_ When I do a "!" in Marpa it will have an exact theoretical description -- not just an example -- and will work in all case.
16:48 gleki But we are in CFG so what can be the solution here?
16:48 idiosyncrat_ s/all case/all cases/
16:48 gleki idiosyncrat_: you have such plans???
16:48 idiosyncrat_ Not short time, but I think about it often.
16:49 idiosyncrat_ s/short time/short term/
16:49 gleki idiosyncrat_: so you are saying my example with (in regex) "a[b-z]" but not "aa" is not possible in Marpa?
16:50 gleki can you recommend other parsers then but those that support left recursion?
16:51 idiosyncrat_ It might be possible but, no, there's no general "!" construct, and there isn't a reliable one in PEG either.
16:52 idiosyncrat_ But I think your example could be done as you've written it: <a> <not a>
16:53 gleki Actually I'm more interested in an extension to regex with exclusion of "([a-z])\1"
16:53 gleki but well, yeah, it's what you said
16:53 gleki ignore my last line about regex
16:54 gleki And aredridel's javascript parser is so short in code. i wonder if it can be fixed to support negative lookahead but im a complete noob.
17:13 maybekoo2 joined #marpa
21:40 idiosyncrat_ joined #marpa
22:09 shadowpaste joined #marpa
23:55 idiosyncrat__ joined #marpa
23:58 idiosyncrat_ joined #marpa

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary