Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-07-26

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

All times shown according to UTC.

Time Nick Message
00:21 djns I was reading about unrestricted grammars. Trying to understand what the function of multiple items on the LHS would be. This is not well documented on the Internet
00:36 idiosyncrat There's a book by Grune & Jacobs which covers the parsing of CSG's -- It's first edition IIRC is online
00:38 idiosyncrat With Marpa, I've obviously committed to doing CFG's.  A good CFG (general BNF) parser like Marpa, plus hackery, may be the best approach to CSG's ...
00:39 idiosyncrat or another algorithm, aimed at one of the larger grammar classes, may turn out to be the best approach.
00:40 idiosyncrat I know some people have nibbled at CSG's using Marpa, but I don't think I ever heard of the results, if any.
00:54 djns i suppose that: "4" <rule1>::= "1" "2"; <start>::="4" <rule1> "3"; produces: <start>::="1" "2" "3";
01:08 idiosyncrat Yup.
01:19 djns Would those rules match a string 123, if used in a lexer, since the effective rule is <start>="1" "2" "3";
01:19 djns It sort of looks like the rules modify themselves?
01:53 idiosyncrat joined #marpa
01:54 idiosyncrat Yes, as I read your example, that CSG accepts the string "123"
02:23 idiosyncrat joined #marpa
02:24 idiosyncrat Probably will convert my Toshiba laptop to Linux, starting with booting from a separate partition
02:24 idiosyncrat What distros do folks use/recommend?
02:44 djns I liked Ubuntu actually until they decided to fracture Linux with Mir
02:46 djns kde+ubuntu
02:54 jeffreykegler joined #marpa
03:08 ronsavage joined #marpa
03:29 ronsavage jk: I use Debian stable. Jean-Damien uses Debian Unstable. Between them lies Debian Testing. I go for Stable because - for me - Reliability comes first, middle and last.A huge majority (sic) of the servers in the internet use Debian stable. I.e. not end-user machines like your PC or mine, but the servers.
03:50 jeffreykegler djns, ronsavage: Thanks
03:51 jeffreykegler I guess I'll install Debian stable intially.  I can always decide to go bleeding-edge like Jean-Damien later.
05:01 koo7 joined #marpa
06:38 ronsavage jk: Exactly.
06:40 ronsavage I've finally found time to update my website so it has Data::RenderAsTree V 1.03 and MarpaX::Languages::Lua::Parser. I've also just submitted the latter to CPAN, since it's new pre-req is the former.
07:42 lwa joined #marpa
10:47 koo7 joined #marpa
10:48 mvuets i thought subsequent G::value() calls would give me all possible parsed trees (when parsing is ambiguous)
10:48 mvuets ...ah! it does, unless i invoke ambiguous()
10:49 mvuets ok, it's even well documented (-:
10:53 mvuets oh boy, Marpa is so facinating! *_*
11:03 pczarn joined #marpa
11:43 mvuets whoo-hoo! so this sorta grammar: S ::= Subject Predicate;  Predicate ::= Verb | Object
11:43 mvuets ...ambiguously parses 'mi moku' into two trees: (S (Subject 'mi') (Predicate (Verb 'moku')))
11:43 mvuets and: (S (Subject 'mi') (Predicate (Object 'moku')))
11:43 mvuets Marpa++
11:48 mvuets i really liked the introspection/reflection Marpa capabilities such as $Marpa::R2::Context, rule_expand(), and symbol_display_form()
12:15 pczarn Which algorithm for building SPPFs is used in Libmarpa? The lazy one (Buildtree) or the eager one (integrated)?
12:17 pczarn is either one strictly better?
14:25 djns joined #marpa
14:33 CQ joined #marpa
14:46 djns hi
15:05 djns hi
15:27 mvuets > Unproductive symbol: CompoundSubj
15:28 mvuets how to get more insights into this warning?
15:33 djns *shrugs* maybe someone else knows
15:33 pczarn mvuets: I think one of your symbols is unused
15:35 pczarn mvuets: or rather one of your rules is unproductive: http://search.cpan.org/dist/Marpa-R2/pod/Vocabulary.pod#Useless_rules
15:36 idiosyncrat joined #marpa
15:38 idiosyncrat pczarn: re http://irclog.perlgeek.de/marpa/2015-07-26#i_10956264 -- I think my approach to constructing the SPPF (= bocage) would be called eager.
15:38 idiosyncrat I think it's an open question which is best.
15:39 idiosyncrat pczarn: re http://irclog.perlgeek.de/marpa/2015-07-26#i_10956995 -- Good catch!
15:41 pczarn so I have the same approach implemented :), with Leo optimization disabled so far
15:41 mvuets pczarn++
15:47 mvuets http://paste.scsys.co.uk/495364 - isn't it great?! (o:
15:51 idiosyncrat mvuets: I does look nice. :-)
15:51 idiosyncrat s/I/It/
15:54 idiosyncrat mvuets: There is also http://search.cpan.org/~jkegl/Marpa-R2/pod/ASF.pod
15:55 idiosyncrat It's a way of traversing the tree top-down, which can allow you to efficiently prune the possibilities.
15:57 * idiosyncrat was thinking of trying to imitate the Steve Jobs routine, where he'd announce a set of things, make it look like he had just done the Grand Finale, and then would say, "Oh, yes.  And now for something really cool!"
15:57 pczarn mvuets++; building Markov chains is great too
15:58 * idiosyncrat was worried about his delivery, though
16:40 idiosyncrat joined #marpa
16:40 idiosyncrat mveuts: I hope there will be video of your talk tomorrow.
17:14 mvuets idiosyncrat: aye, that ASF pod page inspired me to produce the aforementioned output. ASF itself looks rather complicated to me now
17:15 mvuets pczarn: unfortunatelly i think i'm not going to make it to Markov chains this time. too little time left, and still too much to do. but if everything works out well, i'll be giving a talk on the same topic again, so could try to refine and extend it
17:16 mvuets idiosyncrat: talks are usually recorded --- videos are not always published though (-:
17:46 idiosyncrat joined #marpa
19:44 mvuets joined #marpa
19:46 mvuets is there a tell-bot on this channel?
20:13 mvuets BTW, re that my question on 'Unproductive symbol: CompoundSubj', here are two relevant rules from the grammar:
20:13 mvuets SubjPred ::= mi Pred | sina Pred | NP_NoMiSina li Pred | CompoundSubj li Pred
20:13 mvuets CompoundSubj ::= NP en CompoundSubj
20:14 mvuets smells like infinite recursion, doesn't it?
20:15 mvuets must be a mistake that slipped away from the author of the original grammar. and Marpa++ caught it and handled gratefully
20:35 koo7 joined #marpa
22:38 ronsavage joined #marpa
23:10 idiosyncrat joined #marpa

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