Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-01-07

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

All times shown according to UTC.

Time Nick Message
00:07 ronsavage joined #marpa
00:12 jeffreykegler joined #marpa
00:12 jeffreykegler re AFL: perhaps it's just something we should keep in mind.
00:15 jeffreykegler Note that the Earley sets in Libmarpa are a large, complex "compact data format" -- the major difference between Libmarpa's internal data and a bytecode would be that it represents potential parses, rather than a sequence of instructions.
01:32 jeffreykegler My doings:
01:33 jeffreykegler Over the past few days I've caught up with various backlogs -- the recent developer's release of Marpa::R2 is one result.
01:33 jeffreykegler Now I'm back to working on the document describing the partial parses technique.  It's taking a bit of time ...
01:34 jeffreykegler not because there are actual problems with the technique, but because I want to document *why* the algorithm is correct, ...
01:34 jeffreykegler which means proofs.
01:35 jeffreykegler Not of everything, but definitely of those things which are of the sort where I say to myself, "I know this is true, but I can't remember why".
01:35 jeffreykegler So I write up the "why?" and that in high falutin't terms is what's called a proof.
01:36 jeffreykegler * falutin't -> falutin'
01:38 jeffreykegler Another part of this is coming up with terminology, and I plan to borrow some from biology -- the partial parses use special lexemes and non-terminals to match up at the edges ...
01:38 jeffreykegler and I will call these special symbols "nucleobases" and "nucleotides" by analogy with DNA ...
01:40 jeffreykegler My method for putting the partial parse trees together can be thought of as kind of like DNA replication ...
01:41 jeffreykegler A left edge has symbols (nucleotides) and these are matched to symbols to create a right edge and to "transcribe" the left edge of the old parse into the right edge of the new one.
01:42 jeffreykegler Once both parses exist, the nucleotide symbols are matched up, and used to join the two parses together.
01:42 jeffreykegler Anyway, going back to writing that up is what I'm doing at the moment.
01:42 sirdancealot joined #marpa
03:01 ronsavage joined #marpa
06:29 ronsavage joined #marpa
06:41 ronsavage Now, I'm going to play with a proof-of-concept grammar for POD.
08:31 sirdancealot joined #marpa
09:44 basiliscos joined #marpa
09:51 lwa joined #marpa
15:46 jeffreykegler joined #marpa
15:50 jeffreykegler ronsavage: Between you and Jean-Damien we should soon have a Marpa-driven parse of the Voynich manuscript. :-)
16:50 koo6 joined #marpa
17:35 basiliscos joined #marpa
17:53 basiliscos joined #marpa
18:09 jeffreykegler A question on perlmonks: "Parsing a config file with braces and nested braces -- http://www.perlmonks.org/?node_id=1112435
18:24 jeffreykegler Continuing the DNA analogy, I may rename what I've been calling several things, including "partial parsing" and "constant space parsing".
18:25 jeffreykegler I am thinking of calling it "strand parsing", on the idea that the partial trees are "strands" which the technique "transcribes" and joins together.
18:27 jeffreykegler The DNA analogy has the disadvantage of importing a few high-falutin' biochemical, which could be seen as pretentious, ...
18:28 jeffreykegler but I think it may have a real advantage -- many people intuitively the idea of the DNA code, and DNA spliting, joining and transcription ...
18:28 jeffreykegler and it supplies a "mental picture" into which the details of my algorithm can fit.
19:36 sirdancealot joined #marpa
19:52 ronsavage joined #marpa
19:56 ronsavage Ah, yes, 'the Voynich manuscript'. I wrote a Marpa parser for that, but stored it in /tmp, and lost when I re-booted. Hahaha.
19:58 ronsavage As for the DNA connexion. I'm thinking your plan more resembles a zipper, and regrettably, many people are still very resistant to DNA and its connotations of evolution.
19:59 jeffreykegler I did think of sewing, seams and zippers, but in the case of DNA, you have the additional element that the two halves match up in a way that preserves information, which is crucial.
20:00 ronsavage As for http://irclog.perlgeek.de/marpa/2015-01-07#i_9901243, I'll write a demo now.
20:02 jeffreykegler As for the anti-evolution element, my use of the analogy only involves the chemistry -- at the moment I would hope that the parses do *not* evolve through random mutation.
20:08 ronsavage Hmmm. Zips work precisely because the 2 sides match, but yes, the preservation-of-information argument is good. Your call
20:23 ronsavage After a few test runs, I don't think Text :: Balanced :: Marpa is a good fit for reading config files.
20:48 flaviu joined #marpa
21:03 flaviu joined #marpa

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