Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-07-12

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

All times shown according to UTC.

Time Nick Message
01:22 jeffreykegler joined #marpa
01:23 jeffreykegler If the Kollos thing works out, it will probably mean that Marpa::R2 becomes stabilized -- no or only small new features, just bug fixes.
01:24 jeffreykegler Given the kind of user base Marpa::R2 is picking up, it is probably a good thing to move the focus from breaking new ground to supporting those users it has.
01:26 jeffreykegler There's a lot of room for users of Marpa::R2 to innovate, just with the features it has, and added stability may well be a plus for that.
04:51 jeffreykegler Just watched http://vimeo.com/1485123 -- Roberto Ierusalimschy (one of the creators of Lua) describing a PEG parser he wrote in Lua
04:53 jeffreykegler I was watching it for ideas about how to implement a parser in Lua, not because of any interest in PEG.
04:54 jeffreykegler Particular I wondered how he did the syntax -- note he implemented a pattern matcher, *not* a parser and the syntax you want in the two are substantially different.
12:07 shadowpaste joined #marpa
12:07 hobbs joined #marpa
15:21 jeffreykegler joined #marpa
16:07 jeffreykegler By the way, though it may not immediately be recognizable as such, Kollos (Marpa+Lua) is an initiative for a new language, though of a very different kind from the usual.
16:08 jeffreykegler I see a future of small, highly specialized and evolvable, malleable languages -- languages which can so quickly morph that family resemblances are extremely hard to spot.
16:09 jeffreykegler Past language initiatives have been about "one language to solve it all".
16:10 jeffreykegler That's because the parser for a useful language was such trouble to create, and so hard to change, you were forced to get as much out of it as you could.
16:11 jeffreykegler So that even avowedly special purpose languages still sought to have large, general problem domains.
16:13 jeffreykegler Kollos is a language initiative without a syntax specification, a typing system, a semantics or a problem domain.
16:13 jeffreykegler It's a language initiative which is about a language-building toolset.
16:43 jdurand joined #marpa
16:44 jdurand Re http://irclog.perlgeek.de/​marpa/2014-07-12#i_9014095 - may I understand the possiblity to create its own keywords - a bit like Moose@perl is doing with its "sugar"s
16:45 jeffreykegler jdurand: Going well beyond that.
16:46 jeffreykegler Kollos would be to languages what Lua is to games.
16:49 jeffreykegler So you could link into Kollos a VM, a lexer, write your grammar, put them together and you've got your new language.
16:50 jdurand ok I see - good - those wanting to particpate shall investigate in what - isn't lua API out of context with your latest sentence
16:52 jeffreykegler In a sense, yes.  Lua is the fall-back, basic VM and semantics.
16:53 jdurand Kollos should be a sort of turing complete parser if I understand well
16:53 jeffreykegler I would not state it quite that way.
16:54 jeffreykegler Lua extends the *semantics* build into Marpa to be Turing complete.
16:54 jeffreykegler But I suppose with procedural Lua-powered parsing, you could also say that the languages could not be Turing-complete as well.
16:55 jeffreykegler That is, right now in the SLIF DSL, you could not write a Turing machine -- you'd have to specify an action, and use Perl code.
16:56 jeffreykegler (And that limitation was a deliberate choice.)
16:57 jeffreykegler In Kollos, your actions Lua will be Lua code, and Lua is Turing-complete.
16:57 jdurand ok understood
16:59 jeffreykegler I am still pretending that I have not finally made the decision to go with Lua. :-)
17:00 jeffreykegler But extensibility, embeddability and ability to stay out of the way when need to be == those 3 have emerged as the main criteria.  And Lua meets them like nothing else.
17:01 jdurand coming back to my first question, is there anything others than you (the core developer) can do
17:01 jeffreykegler Absolutely!
17:02 jeffreykegler For now, others learning more about Lua would be useful -- I'm very new coming to it.
17:03 jeffreykegler If someone on this channel starts reading about & playing with Lua now, once I get going, if I do something stupid, you might know enough to point that out. :-)
17:03 jeffreykegler In all seriousness, that sort of thing is a big help.
17:04 jdurand ok - I commit to investigate into lua then
17:04 jeffreykegler My hope is that a Lua framework for Marpa will be something others can play an active and even leading role with -- something which is not currently possible in the libmarpa/XS/Perl framework.
17:05 jeffreykegler As one example, you could do a THIF in Lua.
17:06 jeffreykegler Kollos might start there.
17:07 jeffreykegler And even if I do a fresh start, the experience will be helpful.
17:08 jeffreykegler I am still in the road-mapping stage, but I think a THIF in Lua is almost inevitably the first stage.
17:08 jeffreykegler jdurand: by the way, on the topic of Debian ...
17:08 jdurand yep-  ok - this will be exciting, also, to extend my languages/framework knowledge
17:09 jeffreykegler do you have a way of telling how far they are along with Marpa::R2 and libmarpa?
17:11 jeffreykegler As another suggestion of an area to help, I will be looking at CPAN's Lua packages.
17:11 jeffreykegler Advice as to which might be best would be very helpful.
17:14 jeffreykegler Correction to http://irclog.perlgeek.de/​marpa/2014-07-12#i_9014266 -- "could not be Turing-complete" -> "could be Turing-complete"
17:18 jdurand Re debian - none will look tonight
17:18 jdurand Re lua - will look a lua general and dev. doc first I think
17:19 jeffreykegler jdurand: I'd promised them some changes, which were made in the latest Marpa::R2 indexed release, so now I *think* the ball is back in their court.
17:20 jdurand yes. looked to http://anonscm.debian.org/gitweb/?p=pkg-pe​rl/packages/libmarpa-r2-perl.git;a=summary. No official commit since 10 days it seems.
17:22 jeffreykegler jdurand: Thanks.  Jonas was traveling -- he might not have been back long enough to catch up with this.
17:24 jdurand Yes -
17:27 jdurand afk
17:27 jeffreykegler jdurand: thanks!
19:52 jeffreykegler1 joined #marpa
22:57 ronsavage joined #marpa

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