Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-01-08

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

All times shown according to UTC.

Time Nick Message
01:01 jeffreykegler joined #marpa
01:45 flaviu joined #marpa
03:45 jeffreykegler joined #marpa
03:56 sirdancealot joined #marpa
04:51 ronsavage joined #marpa
06:54 basiliscos joined #marpa
09:00 basiliscos joined #marpa
11:00 lwa joined #marpa
17:02 jeffreykegler joined #marpa
17:03 jeffreykegler On perlmonks, choroba submitted a solution to the config file parsing problem -- http://www.perlmonks.org/?node_id=1112655
17:10 jeffreykegler A blog post on what the writer is looking for in a static code analyzer -- it refers to my work several times: http://blog.buguroo.com/why-parser-generator-tools-are-mostly-useless-in-static-analysis/
17:33 sirdancealot joined #marpa
17:34 koo6 joined #marpa
18:05 rns joined #marpa
18:10 rns jeffreykegler: Hi, I'm experimenting with using marpa_r_terminals_expected() to reduce the number of matches in the lexer -- https://github.com/rns/libmarpa-bindings/blob/master/lua/json.lua#L346 -- would that be a legitimate use of it? I mean the test (if tiny) passes, but it is ok to use marpa_r_terminals_expected() that way?
18:10 rns s/it is/is it/
18:11 rns Sorry, have to go, AFK.
18:11 rns left #marpa
18:12 jeffreykegler rns: re http://irclog.perlgeek.de/marpa/2015-01-08#i_9906276 -- yes, it is a legitimate use.
18:14 jeffreykegler In fact, in thinking of lexers for Kollos, I have planned to use that technique to both enforce LATM, and to cut down on the cost of matching.
18:14 * jeffreykegler assumes that rns will backlog
18:36 jdurand_ joined #marpa
18:36 jdurand_ Re http://irclog.perlgeek.de/marpa/2015-01-08#i_9906307 - indeed marpaWrapper is doing that as well in the LATM mode
19:18 jdurand_ Re https://groups.google.com/d/msg/marpa-parser/ZIZ003_GUig/3ti4II8HRrgJ - Jeffrey contacted me off-the-group to suggest it would be nice to have comment information in a parsing tool, and I fully agree
19:19 jdurand_ I suggested him to create an internal event "'discard", alike the internal event "'exhausted", triggered if a corresponding option discard => 'event' would be setted in the recognizer's construct.
19:19 jdurand_ This hypothesis is open and we would be glad to have other's opinions
19:20 jdurand_ Now Jeffrey, if you don't mind, let me go further a bit, despite Marpa::R2 is sort of frozen: you remember that the L0 grammar does support event, pause, etc... adverbs but they do nothing.
19:21 jeffreykegler An implication would be that I back-burner Kollos (again) to add discard events to Marpa::R2, and I'd like to know folks thoughts on this.
19:21 jdurand_ Allowing a :start symbol to be the :discard symbol could perhaps automagically enable the user of these adverbs in part of L0 ? This is a nice-to-have, I would be at least happy to see the "'discard" internal event and only this.
19:22 jdurand_ Yes. If people are not interested except me, then "tant-pis"
19:23 jeffreykegler Actually, the vote counting is very idiosyncratic -- I might go ahead if only Jean-Damien wants this, because I suspect he won't be the last one.
19:25 jeffreykegler Re: "a :start symbol to be the :discard symbol", I didn't really understand what this meant -- that is, I didn't understand what the proposed syntax would look like.
19:25 jeffreykegler But I don't want others feedback on this.
19:31 jeffreykegler Oops!!!! CORRECTION: I *do* want others feedback on this.
19:32 jeffreykegler "This" meaning going back to Marpa::R2 and adding "discard events"
19:36 jdurand_ Fine with me. Let's way other's feedback. Btw, see my answer on the group, perhaps the "priority" of this "'discard" event request could drop down significantly, depending on Zeev's feedback
19:38 jeffreykegler Maybe, but these things can take on a life of their own.  Note that my work on "strand parsing" is a follow-up to a question by flaviu, but it won't produce a useful result in his timeframe, and he may at this point even have forgotten the question.
19:44 basiliscos joined #marpa
19:46 pczarn joined #marpa
20:05 jdurand_ Understood.
20:18 lwa Re: discard events – I think that falls into the category of “convenient, but no game-changer”. If I wanted to, I could write a grammar to the same effect that does not :discard comments, at the cost of a messier grammar and more memory use for Marpa. In particular, the latter issue would be solved if partial parses were introduced.
20:19 lwa But until then, discard events seem to have a very good implementation cost to usefulness ratio, so I would vote for such an immediately useful feature over long-term projects. I'd probably use it in a few of my projects, but I do not really need this feature.
20:48 jeffreykegler lwa: Thanks!
20:50 jeffreykegler My initial reaction to was that, yes, jdurand should just rewrite his grammars.  But then I realized this isn't all that desirable when you've got a large grammar, carefully honed to closely follow several standards ...
20:51 jeffreykegler A rewrite to eliminate discards would really mess a grammar like that up.
20:53 jeffreykegler AFK
21:00 ronsavage Re discard => 'event', is the idea to have something like :discard ~ whitespace pause => before event => discard? And yes, I did /not/ write discard => event! I'm deliberately making it look like :lexeme..... I can think of use case. ATM GraphViz2::Marpa discards comments (it has to), but there is an reasonable argument that they be preserved, since they will (presumably) carry meaning to the author. I like the idea, especially since the 'paus
21:00 ronsavage Grrrr. I said ': discard'. Just delete the space.
21:40 ronsavage joined #marpa
22:04 basiliscos1 joined #marpa
23:01 ronsavage I've updated the FAQ so that links in the TOC (Table of Contents) jump to the questions within the FAQ: http://savage.net.au/Perl-modules/html/marpa.papers/chapter3.html.
23:01 ronsavage Also, there are new questions: 6, 25, 26.
23:19 basiliscos joined #marpa
23:34 basiliscos1 joined #marpa
23:42 basiliscos joined #marpa

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