Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-08-02

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

All times shown according to UTC.

Time Nick Message
00:31 ronsavage Aria: Your page http://dinhe.net/~aredridel/.notmine/PDFs/Parsing is off the air!
00:32 ronsavage jk: Rehttp://irclog.perlgeek.de/marpa/2015-08-01#i_10989943. It's 10:30 am here. By lunchtime I hope to release an updated version of *::Lua::Parser.
01:00 Aria Whoops! Should be working, not sure what happened.
01:01 Aria In the mean time, a friend pointed me at www.cse.chalmers.se/~bernardy/PP.pdf
01:13 Aria Alright. That's working again
01:15 ronsavage joined #marpa
01:15 ronsavage Aria: Yes, I see it now. Thanx.
01:35 ronsavage I've incorporated and pushed patches from https://github.com/ronsavage/MarpaX-Languages-Lua-Parser/issues/2, but am getting "The Lua source is ambiguous: Length of symbol "<stat item>" at line 153, column 1 is ambiguous" running it on lua.sources/constructs.lua.
01:48 ilbot3 joined #marpa
01:48 Topic for #marpa is now Start here: http://savage.net.au/Marpa.html - Pastebin: http://scsys.co.uk:8002/marpa - Jeffrey's Marpa site: http://jeffreykegler.github.io/Marpa-web-site/ - IRC log: http://irclog.perlgeek.de/marpa/today
03:14 purmou ooh Aria, how long have you been collecting those PDFs? I wish I found that earlier heh
03:14 Aria Going on a year now
03:15 purmou do you have one that describes in detail the bit vector approach?
03:30 Aria Nope.
03:31 Aria That I got from marpa.
03:31 Aria Plus my own spinning through it a few times.
04:54 jeffreykegler joined #marpa
04:55 jeffreykegler purmou: bit vectors are used in various places
04:55 jeffreykegler the most important is computing transitive closures, things like
04:55 jeffreykegler I know symbol A directly reaches B, D reached F, etc.
04:55 jeffreykegler s/reached/reaches/
04:56 jeffreykegler What symbols indirectly reach each other?
04:57 jeffreykegler This winds up being useful/necessary for finding cycles, productive symbols, accessible symbols, etc., etc., etc.
04:58 jeffreykegler There a textbook by Grune and Jacobs which is kind of the standard parsing reference for those of us using non-mainstream grammars ...
04:58 jeffreykegler and it has nice coverage of this in its early chapters.
04:59 jeffreykegler There's also a Wikipedia page: https://en.wikipedia.org/wiki/Transitive_closure
04:59 jeffreykegler Though it's very math-y and looking at it you'd have no idea the concept is useful in parsing theory.
05:00 jeffreykegler Anything I need a transitive closure for, I basically wind up converting to a bit matrix and using Warshall's algorithm.
05:01 jeffreykegler Good night, all!
06:38 pczarn joined #marpa
08:45 ronsavage joined #marpa
10:30 lwa joined #marpa
14:18 koo7 joined #marpa
14:29 djns joined #marpa
14:42 pczarn joined #marpa
15:33 rns ronsavage: re http://irclog.perlgeek.de/marpa/2015-08-02#i_10990713 -- Github issue filed -- https://github.com/ronsavage/MarpaX-Languages-Lua-Parser/issues/3
15:55 rns jeffreykegler: re http://irclog.perlgeek.de/marpa/2015-08-01#i_10989943 -- actually, yes: you can tell about libmarpa and kollos -- some people in the lua community are good at C, so they might put it to good use; Ron can tell about Lua::Parser; me -- yes, I'm not ready to tell the lua community about Lua::AST yet.
17:00 jeffreykegler joined #marpa
17:01 jeffreykegler I've sent an email introducing Marpa and the Kollos project to Lua mailing list
17:01 jeffreykegler http://news.gmane.org/gmane.comp.lang.lua.general/cutoff=118285
17:05 koo7 joined #marpa
17:09 jeffreykegler joined #marpa
17:28 purmou joined #marpa
19:24 jeffreykegler joined #marpa
20:27 djns So Kollos will be accessible to both Perl and to Lua?
20:35 pczarn djns: initially only in Lua, I believe
20:35 jeffreykegler Both Perl and Lua
20:35 jeffreykegler In the email to the Lua group I emphasized what it means for them.
20:36 jeffreykegler Basically, as the title says, "Kollos = Marpa + Lua"
20:36 jeffreykegler Lua becomes *part* of Marpa in Marpa's next rev.
20:36 djns Yes.
20:37 jeffreykegler Marpa::R3 will interface to Kollos -- from the Perl point of view things will be much as before, except Marpa is now enhanced ...
20:37 djns good
20:37 jeffreykegler the SLIF is replaced with the LUIF ...
20:38 jeffreykegler and Marpa::R3's built-in semantics (which was never Perl) is now Lua.
20:39 jeffreykegler I expect people writing semantics for Marpa::R3 will now have their choice: Perl or Lua.
20:39 jeffreykegler Lua will be much faster, since it's native.
20:40 jeffreykegler On the other hand, Lua's ruthless minimalism (for example, no real ternary operator -- x ? y : z) will make Perl the preference for many ...
20:41 jeffreykegler and by being clever you can get the best of both.
20:42 jeffreykegler From the Perl point of view the only "set back" is that Perl loses the contest for being Marpa's built-in semantics language ...
20:43 jeffreykegler But this was not the kind of contest which Perl 5 (never mind Perl 6) could ever win, or really even tries to win.
20:44 pczarn does the lack of inline actions help somehow with transition?
20:44 jeffreykegler That is, Perl does not try AFAICT to be your "easy to drop in" extension language.
20:44 jeffreykegler pczarn: huh?
20:44 jeffreykegler Oh, you mean the fact that I never did create a full semantics language for Marpa::R2?
20:45 pczarn yes, since the SLIF is being replaced
20:46 jeffreykegler pczarn: Not sure how to answer that -- I'm not trying to maintain backward compatibility from LUIF to SLIF in any case.
20:46 jeffreykegler I suppose psychologically you could say the lack of inline actions makes the transition easier ...
20:47 jeffreykegler because if that language had been Perl, there might be a sense that I was moving away from Perl.
20:47 jeffreykegler The reason I never went with inline semantics is that I wanted to keep my options open ...
20:48 jeffreykegler and also inline-or-not, Perl semantics would be high-cost, and high-cost stuff should look high cost.
20:48 jeffreykegler That is, syntax like
20:48 jeffreykegler A ::= B C { $B + $C }
20:49 pczarn ah, sure, conservative decisions are good, too
20:49 jeffreykegler makes it look the action is a pretty trivial thing when actually it is extremely complex to implement and a real performance bottleneck in use.
20:50 jeffreykegler The use of Lua could be said to be Perl *6* friendly ...
20:51 koo7 joined #marpa
20:52 jeffreykegler because porting a C89 based package (and basically if I include Lua, I'm still strictly C89 in all portability senses) ...
20:52 jeffreykegler is a lot easier I expect than porting a lot of complex Perl 5 XS logic to its equivalent in Perl 6.
21:40 koo8 joined #marpa
21:46 koo9 joined #marpa
22:04 koo9 joined #marpa
22:20 koo9 joined #marpa
23:42 ronsavage rns, jk: Re: http://irclog.perlgeek.de/marpa/2015-08-02#i_10991988. I've released V 1.04 of *::Lua::Parser. It's on CPAN already, and github, and my website. Thanx!
23:43 ronsavage And on that, we /really/ need a pretty-printer for Lua.........

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