Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-03-12

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

All times shown according to UTC.

Time Nick Message
00:00 koo5 joined #marpa
00:44 jeffreykegler joined #marpa
00:44 flaviu joined #marpa
00:51 jeffreykegler http://jeffreykegler.github.io/Ocean-of-Awa​reness-blog/individual/2015/03/linear.html
00:51 jeffreykegler A new blog post, addressing skepticism re Marpa's time complexity claims.
01:01 flaviu joined #marpa
03:44 Aria https://lobste.rs/s/sumfav/linear_yeah_right
04:39 ronsavage joined #marpa
05:06 btyler_ joined #marpa
06:54 rns joined #marpa
06:58 rns A good read -- stating that Marpa's complexity claims have theoretical proof in the prior art and then inviting sceptics to obtain a proof by experiment looks like a good idea.
06:58 rns left #marpa
07:06 rns joined #marpa
12:29 koo5 joined #marpa
13:27 lwa joined #marpa
18:00 jeffreykegler joined #marpa
19:23 jeffreykegler My last 2 blog posts have been in the class of "things I probably should get off my desk and write up carefully"
19:23 jeffreykegler Even if they weren't obviously going to be big crowd-pleasers.
19:24 jeffreykegler That is, I expect they will be the answers to frequently asked questions.
19:24 jeffreykegler Along this lines, I am thinking of blog posts on the topic "When in Marpa linear?"
19:24 jeffreykegler That is, how can you tell if your grammar will be parsed by Marpa in linear time.
19:25 * jeffreykegler cannot eyeball a grammar and tell whether it is LR-regular or not, and expects most other programmers cannot either.
19:27 jeffreykegler This series of posts would explain Marpa's linear class of grammar in simpler terms, so that you can eyeball the BNF and have a good idea.
19:27 jeffreykegler These posts would take time away from strand parsing and Kollos, and I wonder what the consensus on this channel is about doing that.
19:37 koo7 i know i could start plugging kollos into a gui for https://github.com/opennars/opennars/ if it existed:)
19:38 koo7 and that would introduce it to a bunch of smart folks
19:41 koo7 plug in perl? not so realistic, or only as an alternative
19:43 koo7 smart folks who just want to get some parsing done, and will never dig into the esoteric corner case details of pegs vs whatever
19:43 koo7 they are probably using regexes right now
19:48 koo5 joined #marpa
20:08 rns koo7: so those guys would be interested in a java interface to Marpa?
20:12 rns jeffreykegler: re http://irclog.perlgeek.de/m​arpa/2015-03-12#i_10267855 -- this can be useful -- I remember reading "What about Marpa?" section on a similar topic in http://jeffreykegler.github.io/Ocean-of-Aware​ness-blog/individual/2014/11/successful.html
20:13 jeffreykegler rns: Yes, I semi-withdrew that post, because it was in effect two posts in one, neither successful.
20:14 jeffreykegler A post on what makes a parser successful, where I felt I did not get my point across well.
20:15 jeffreykegler And 2nd the stuff about linearity, which I backed myself into trying to explain the first part, but had to compress it so much it was well-nigh incomprehensible
20:16 koo5 rns: well the parsing part of the story is my cup. If you mean java bindings to libmarpa, yes but i cant promise we'll get as far as to make use of them, im thinking about starting simple, with an UI for which something like SLIF would suffice.
20:16 jeffreykegler I have errands to run.  Will backlog.
20:17 jeffreykegler AFK
20:17 rns jeffreykegler: When I was reading it I was thinking -- perhaps there can be a set of tests on a grammar to tell if it can't be parsed by Marpa in linear time so that to write a utility (gramlint) to check for them.
20:19 rns koo5: Well java bindings to libmarpa can be relatively easy with libffi (that would give pure BNF), but re-implementing SLIF in java is more difficult.
20:19 rns Or you mean you can use Marpa::R2's SLIF directly?
20:22 koo5 yes im aware, bindings will be easy as they were for python hopefully
20:23 koo5 personally im not gonna reimplement anything
20:23 koo5 use R2...maybe, but now you say that...perl6 has MoarVM?
20:24 koo5 and doesnt perl6 have some backwards compatibility switch?
20:25 rns There has been a p5 to p6 translator, don't really know about compat options.
20:27 rns Basically, if you have a libmarpa binding, you can add a regexp based lexer and table-based grammar and have a basic parser, e.g. https://github.com/rns/libmarpa-b​indings/blob/master/lua/json.lua
20:28 koo5 There are multiple projects aiming to make it possible to use Perl 5 modules directly from Perl 6 code:v5 A slang that lets Rakudo itself parse blocks of Perl 5 code, and compile them to the same bytecode that it compiles Perl 6 code to.Inline::Perl5 Uses an embedded instance of the perl interpreter to run Perl 5 code called from your Perl 6 script.Of these, Inline::Perl5 is currently the furthest along and the most promising.
20:30 rns Ok, but that probably doesn't cover XS modules. Or does it?
20:31 koo5 no idea
20:31 rns same thing here. :)
20:32 koo5 thanks, that lua code will be handy to look at
20:34 koo5 As you know ive done some high-levelizing above libmarpa in the context of my editor, i havent mastered it all yet, like no rule ranks yet. And ive never used the high level interfaces or any other parser engine, I still am not sure what functionality, like some bnf rewriting or something, is there and why exactly
20:36 koo5 The goal is to be able to run all pure-Perl 5 code besides Perl 6. One benefit of this approach (reimplementing Perl 5 with Perl 6's object system) is to be able to pass data/objects easily between both languages. XS will not be supported, since it depends of Perl 5's internal structure.
20:36 koo5 v5.pm for rakudo
20:37 koo5 Inline-Perl5:
20:37 koo5 Module for executing Perl 5 code and accessing Perl 5 modules from Perl 6.
20:37 koo5 Supports Perl 5 modules including XS modules.
20:40 rns using perl5 XS modules from perl6 would be cool and Inline-Perl5 seems to do that -- https://github.com/niner/Inline-Perl5
20:45 rns re http://irclog.perlgeek.de/m​arpa/2015-03-12#i_10268396 -- above a libmarpa binding (Thin interface in the case of Marpa::R2), in the case of SLIF, AFAIK, there is a lexer supporting literals and character classes and grammar as a string.
20:47 rns SLIF grammar syntax is easier to use than BNF and provides lots of useful extensions (generalized precedence, lexical rules, parens to hide symbols from semantics to name a few).
20:48 rns BNF rewriting is done internally, other features (events, rule ranks) are used via Thin interface with supporting XS code.
21:07 rns left #marpa
21:09 ronsavage joined #marpa
22:12 hobbs http://www.reddit.com/r/dailyprogra​mmer/comments/2yquvm/20150311_chall​enge_205_intermediate_rpn/cpcy8fa -- toy problem but it came out pretty nicely :)
22:16 hobbs (as a side note, would it break anything to allow Names::With::Separators in the argument to the bless adverb?)
23:13 jeffreykegler joined #marpa

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