Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-05-05

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

All times shown according to UTC.

Time Nick Message
01:23 idiosyncrat joined #marpa
01:24 idiosyncrat Thought I'd describe my current Marpa activities for those interested.  I had a quiet period, and that means when things heat up and my current task is incomplete, I wind up multi-tracking.
01:25 idiosyncrat First priority is the bug latk found.  I will put the "first priority" in favor a "hot pursuit", but Marpa::R2 bugs are always my first priority.
01:26 idiosyncrat Second priority is the higher Kollos level, which rns is driving at this moment.  I try to respond and review so that I don't become a roadblock.
01:27 idiosyncrat Third priority is a tie:
01:27 idiosyncrat Third priority #1 is the lower level of Kollos, so that it's ready when the higher level is ready.
01:28 idiosyncrat Third priority #2 (but the current hot pursuit item) is some theory ideas which will clear up certain points about when Marpa is linear.
01:29 idiosyncrat Third priority #2 is what I'm doing right now, because the ideas may disappear if not written down, but latk's bug will be gotten to shortly.
02:53 ceridwen idiosyncrat, What don't you know about when Marpa is linear?
03:37 ronsavage joined #marpa
03:38 hobbs joined #marpa
04:13 idiosyncrat ceridwen: Actually, for an arbitrary grammar, it is undecidable whether Leo's algorithm (and therefore Marpa) is linear.
04:18 idiosyncrat I'm now working on latk's bug.  I note http://search.cpan.org/dist/Marpa-R2/pod​/Semantics.pod#The_per-parse_constructor
04:19 idiosyncrat where my description of the per-parse constructor is not consistent with the rest of the document -- it says the per-parse constructor *is* called with the per-pasre argument
04:19 idiosyncrat as its only argument
04:20 idiosyncrat which is not only inconsistent with the other Marpa::R2 docs, but inconsistent with the way the new() constructor is usually called --
04:20 idiosyncrat the first argument is usually the class name.
04:21 idiosyncrat To get consistent documentation and even-close-to-intuitive behavior, I'm going to have to break compatibility here.
05:36 idiosyncrat OK.  latk's bug is fixed, and there's a test script ready for addition to the test suite.
05:37 idiosyncrat Next comes a developer's release and a discussion phase, but that's a slower process and at a lower priority.
05:37 idiosyncrat Good night!
06:40 hobbs joined #marpa
09:36 ronsavage joined #marpa
11:04 rns Great article about ambiguity resolution by human parsers -- https://www.academia.edu/12111969/When​_relative_clauses_are_preferred._An_un​expected_garden_path_and_its_consequen​ces_for_sentence_processing_and_syntax
16:01 idiosyncrat joined #marpa
16:38 hobbs joined #marpa
17:18 idiosyncrat re http://irclog.perlgeek.de/m​arpa/2015-05-05#i_10548953 and the resolution of ambiguity in natural languages like English,
17:19 idiosyncrat I've argued the problem is undecidable: http://jeffreykegler.github.io/Ocean-of​-Awareness-blog/individual/2012/03/the-​syntax-of-english-is-undecidable.html
18:40 lucs joined #marpa
19:41 lwa joined #marpa
20:47 hobbs joined #marpa
20:47 hobbs joined #marpa
20:51 idiosyncrat joined #marpa
20:52 idiosyncrat I just uploaded Marpa-R2-2.105_000 to CPAN -- it has the fix to latk's per-parse-argument issue.
20:52 idiosyncrat It's a release candidate for a new major version Marpa::R2 3.0
20:53 idiosyncrat Testing is appreciated!  There is some possibility current apps may be relying on the buggy behavior, which is why the new major version.

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