Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-01-17

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

All times shown according to UTC.

Time Nick Message
00:38 ronsavage I've added https://github.com/jddurand/MarpaX-Role-Parameterized-ResourceIdentifier to Marpa's homepage list of Perl modules.
01:54 Idiosyncrat joined #marpa
01:55 Idiosyncrat re https://medium.com/@nayafia/how-i-stumbled-upon-the-internet-s-biggest-blind-spot-b9aa23618c58#.f1hykxy36
01:55 Idiosyncrat I just reread this -- it's about open source funding.
01:55 Idiosyncrat Very well researched, very well thought out.
01:55 Idiosyncrat Very applicable to Marpa's situation.
01:56 Idiosyncrat No solutions, but a good taut analysis and description of the issue.
01:58 Idiosyncrat The forum she is using, I'm not familiar with it -- medium.com.  Anybody know about them?
02:22 koo7 joined #marpa
02:48 ronsavage joined #marpa
03:24 ronsavage JK: Yes. They host the Physics arXiv blog: https://medium.com/the-physics-arxiv-blog, So, in a sense, they're famous ;-).
03:24 ronsavage See also https://medium.com/life-hacks/the-101-best-genius-life-hacks-that-make-life-easier-86c2152a8f7a#.tdony2tse
05:03 Idiosyncrat re medium.com -- I've been looking for a new site, for when I want to say something less formal than my blog posts,
05:03 Idiosyncrat but too long for twitter.
05:04 Idiosyncrat Google+ seems to be dying, and on FB I have lots of non-technical friends and family, and I try to avoid geeking them out.
05:49 Idiosyncrat Good night!
05:59 kaare_ joined #marpa
13:40 VsyachePuz i completed the reading of C# earley implementation (https://github.com/coonsta/earley) - it is very basic and doesn't contain neither SPPF nor parser for EBNF (which is necessary to start learning/using it in practice)
13:46 ernimril VsyachePuz, "Copyright 2004..." I guess that has not been updated in a long time
13:47 VsyachePuz ernimril: features are important, not the date. Earley's work is from 1969
13:49 ernimril VsyachePuz, sure, his work is from 1969, when is sppf from? did he actually say that implementations has to use ebnf? The source repository you are looking at looks very much like a school assignment "build a parser for a simple calculator"
13:50 ernimril VsyachePuz, use it if you can, extend it or write your own library if you need more.
14:54 VsyachePuz ernimril: I can't implement it myself in the way i wish, because I don't understand, how EBNF relates to grammar with simple rewriting rules (from original earley work)
14:56 VsyachePuz EBNF can contain repetitions and rules should be cut to several records for each alternative
14:58 VsyachePuz wikipedia says that i need create additional rules to convert EBNF to BNF
14:59 VsyachePuz ant this complicates tracing of diagnostics outputs back to original grammar
15:25 VsyachePuz there is a paper http://link.springer.com/chapter/10.1007%2F978-1-4020-3953-9_22
15:25 VsyachePuz but it is not in free access
15:31 VsyachePuz the same, but anohter site (also without text) - https://www.researchgate.net/publication/226022677_An_extension_of_Earley's_algorithm_for_extended_grammars
16:27 VsyachePuz found a russian language paper from 2010 - http://se.math.spbu.ru/SE/diploma/2010/it/Grigoriev_diploma.pdf
16:28 VsyachePuz it propose to represent EBNF rules as Finite-state machine, and encode that state instead of dot into states
16:44 Cheery Someone may want to look at this: https://github.com/cheery/chartparser
16:45 Cheery I plan to later provide an LR parser in similar flavour.
17:55 Idiosyncrat joined #marpa
17:56 Idiosyncrat Re the recent discussion, I think the SLIF is an EBNF variant, in the sense the term ENBF is being used.
17:57 Idiosyncrat But the SLIF code has all sorts of layers, so pulling that logic out would be non-trivial.
17:58 Idiosyncrat EBNF to BNF is best done as a rewrite of the grammar, as opposed to hacks in the parse engine IMHO.
17:58 Idiosyncrat Most papers will be about new parse engines and hacks, because the rewrite is too easy to be publishable.
18:00 Idiosyncrat I'm immersed in documenting the Leo logic, so documenting EBNF-to-BNF is not something I could get to soon.
18:00 Idiosyncrat Perhaps one or more of you could collaborate on it -- it's not that hard to figure out and I can help.
18:01 Idiosyncrat The hardest part is keeping track of the rules and symbols in the original grammar, when you need to do for semantics, error reporting and ...
18:01 VsyachePuz on the page http://marpa-guide.github.io/chapter1.html the term SLIF id used without definition and without link to description of term
18:01 Idiosyncrat (if you implement them) events.
18:02 Idiosyncrat Here's a first hint.  Part of EBNF is usually repetition, that is ...
18:02 Idiosyncrat A ::= B | C | D
18:02 Idiosyncrat The rewrite is
18:02 Idiosyncrat A ::= B
18:02 Idiosyncrat A ::= C
18:02 Idiosyncrat A ::= D
18:03 VsyachePuz yes, i understood the rewriting idea in all three ways - repetitions, optional parts and alternatives
18:03 VsyachePuz but the thought of using FSM state catch me
18:04 Idiosyncrat You mean FSM in the parse engine, as opposed to a preliminary rewrite?
18:04 VsyachePuz yes
18:05 Idiosyncrat If that's your line of investigation, you'll certainly find lots of precedent in the literature, but ...
18:06 Idiosyncrat I've already given my opinion above.
20:40 koo7 joined #marpa
21:50 ronsavage joined #marpa
22:09 Pursuit joined #marpa
22:24 Idiosyncrat joined #marpa
22:50 koo7 joined #marpa
22:58 Pursuit joined #marpa
22:58 koo7 joined #marpa

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