Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-08-11

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

All times shown according to UTC.

Time Nick Message
00:10 Aria Nice.
03:44 ronsavage1 joined #marpa
06:16 ronsavage joined #marpa
16:39 jeffreykegler joined #marpa
16:52 sivoais joined #marpa
17:12 jeffreykegler Larry and I were just talking on the Perl 6 IRC channel: http://irclog.perlgeek.de/​perl6/2014-08-11#i_9167378
17:13 jeffreykegler He'd been interesting in Marpa-powered Perl 6 parser.  Any one interested?
17:14 jeffreykegler Kollos/R3/Lua I think should remain my own focus, otherwise I'd take it on.
17:15 jeffreykegler But I'd certainly be very interested in the project & try to be available for questions, etc.
18:50 jdurand_ joined #marpa
18:50 jdurand_ Re https://groups.google.com/​forum/#!forum/marpa-parser - I do and guess my generation of C code could make a huge point -; in which form is the perl6 grammar
18:54 jeffreykegler jdurand: the perl6 grammar is in a peculiar form -- Perl 6 was intended to be self-parsing and the Perl 6 grammar is that syntax.  IIRC the file is called Std.pm.
18:54 jeffreykegler It would need to be translated into BNF.
18:56 jeffreykegler The link may be https://github.com/perl6/std/ -- before putting in a lot of effort, check on #perl6 that it's really the latest version.
18:57 jeffreykegler My understanding (I may be wrong) is Perl 6 grammars proved too slow for self-parsing, and Std.pm is more of a "standard" than anything else -- though it can be run.
18:58 jeffreykegler The Perl 6 parser is a variation on left parsing, with a notation intended to help with all the custom stuff.
18:58 jeffreykegler This allows control of the backtracking, etc., etc., all stuff which should be ignored IMHO.
18:59 jeffreykegler Almost always it will be hack-ish ways of making a left parser do what Marpa does directly.
19:00 jeffreykegler The file (again we'd need to make sure it is the lastest) is https://github.com/perl6/std/blob/master/STD.pm6
19:03 jeffreykegler If I were doing it, I'd be tempted to use the Synopses, etc, to try to construct BNF, and get a sign-off from Larry and/or the #perl6 channel.
19:03 jdurand_ I see - this is a syntax I am not used with - I wonder how much the semantics interleave with the rules
19:04 jeffreykegler The sign-off would be that the BNF was close enough for a first attempt.
19:04 jeffreykegler Not that it was an exact match.
19:05 jeffreykegler I'd just try to get BNF as a first phase, and a sign-off on that.
19:05 jdurand_ Sure - with a perl5 module on top of that if they do not mind
19:05 jeffreykegler I don't think they will.
19:05 * jdurand_ is wondering is Marpa can be perl6's packaged
19:06 jeffreykegler Huh?
19:06 jdurand_ http://modules.perl6.org/
19:06 jeffreykegler IMHO we should leave that for another day.
19:07 jdurand_ I never understood what can be the difference with what is on CPAN - what is specific to the modules listed here - do you know
19:07 jeffreykegler The package infra-structure for Perl 6 is very new, very much under construction.
19:08 jdurand_ Ok - anyway - since I know nothing with perl6, this cannot be anything else but a perl5 module, up to somebody else to translate - IMHO this is a totally cosmetic thing
19:08 jeffreykegler The sign-off I have in mind is something like: "Yes a Marpa parser for this BNF would be a sufficient proof of concept."
19:09 jdurand_ exactly - proof of concept is enough
19:09 jeffreykegler Certainly for now, yes.
19:10 jeffreykegler Keep in mind that community (like most) thinks of parsing in deterministic / left parsing ways --
19:11 jeffreykegler so there may be real differences of vision.
19:11 jdurand_ Yep I know that -  Ok - just let me say that if somebody else want to start with it right now, do it - I am currently finishing a very novative way to implement DOM3
19:12 jeffreykegler That's why I've been slow about approaching them -- they have been heavily committed to a different (and far more common) way of looking at parsing.
19:12 jeffreykegler Is this what you mentioned on G+?
19:13 * jeffreykegler meant is the DOM3 what was mentioned on G+
19:13 jdurand_ Yes - to myw knowledge only Oracle did that up to now - there is no opensource equivalent approach
19:14 jeffreykegler To be honest, I didn't really follow what was being said there.  I caught the "round trip" idea, and that was about it.
19:14 jdurand_ and I do not understand why... the benefits a huge IMHO: locking/transactional/memory management/storage efficiency are where the DB people are masters - why bother rewrite everything
19:15 jeffreykegler Remind me -- what were the stages?
19:15 jeffreykegler DOM3?  OMG IDL?
19:16 jdurand_ utf8 streaming (done), DOM3 (I brainstormed a lot on it - this is current stage) - SAX/Stax (almost no work: this is just a proxy to Marpa events and recognizer methods, respectively)
19:17 jdurand_ The OMG IDL is a parallel work that I decided to make public because I use that to generate the grammars
19:18 jeffreykegler Re DOM3: I just don't really know enough to have an opinion on the idea, one way or the other.
19:19 jeffreykegler I will have to go AFK soon.
19:20 jdurand_ No pb - note that I asked the question on perl6
19:20 jeffreykegler I had an Internet glitch -- could you repeat the question?
19:20 jdurand_ Just click on http://irclog.perlgeek.de/​perl6/2014-08-11#i_9168175
19:25 jdurand_ ok, thx for your post, I was to say this is to do an attempt with Marpa, you were quicker that I do - looking forward some advice from the perl6 experts in order not to go in the wrong direction
19:36 jeffreykegler jdurand: I'll have to leave soon.  The interaction over at #perl6 may require some patience.  They are very attuned to their current approach ...
19:37 jeffreykegler And remember that for most of the world, what we claim to be able to do is known to be impossible ...
19:37 jdurand_ No pb - many thx
19:37 jeffreykegler so that many will be very dubious about our undertaking.
19:40 jdurand_ That's why this is called a proof of concept - I personnaly do not have at all in my mind something that would mean "we can be do it better" - this is just another approach that I have technical interest to me - and if perl6 can benefit from that, I'll be very happy
19:40 jeffreykegler They are a very friendly, helpful crowd.
19:41 jeffreykegler When you have a start with the Perl6 grammar, maybe you could shadowpaste it on this channel?
19:42 jdurand_ Definitely
19:42 jdurand_ soon -;
19:42 jdurand_ AFK as well

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