Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-10-10

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

All times shown according to UTC.

Time Nick Message
00:47 jeffreykegler joined #marpa
00:47 jeffreykegler I've just completed "final" revs to to the ASF and Glade docs.  These are only of interest to those of you into Marpa's parse forests.
00:49 jeffreykegler I picture Marpa as evolving into a group effort, but I think one duty that cannot be delegated is the reference docs.
00:50 jeffreykegler This is unlike many projects, where you could reasonably say to someone, "Just read the code and write up for each method what it does".
00:51 jeffreykegler With Marpa, there is always a conceptual framework, which is absolutely essential to describe, and which is not self-evident from the code.
01:10 ronsavage I've just completed what I expect is the final draft of the new docs for MarpaX:: Demo :: StringParser, which will ship with all the string parsing code we've dealt with recently.
01:11 jeffreykegler ronsavage: Great!
01:58 flaviu1 joined #marpa
02:03 flaviu1 joined #marpa
02:09 ronsavage That includes the 'conditional preservation of whitespace' logic. I just need to write some tests, and re-create the demo page - the input files got overwritten, but I used git (phew!)
03:19 CQ_ joined #marpa
03:38 jeffreykegler joined #marpa
03:44 ronsavage I've released MarpaX:: Demo :: StringParser V 2.00. I'll have more to say on the Marpa Google group shortly.
04:34 ronsavage joined #marpa
05:30 jeffreykegler joined #marpa
05:31 jeffreykegler I have just uploaded Marpa-R2 2.097_001
05:36 jeffreykegler It's mainly documentation updates.  Testing appreciated!
06:21 ronsavage Marpa::R2 V 2.097001. Test statistics:
06:21 ronsavage Fails: 0. Files: 214. Modules: 6. Passes: 6. Tests: 208.
06:21 ronsavage Duration: 40 seconds
06:29 CQ Marpa::R2 is up to date (2.096000). ...this is from cpan
07:36 rns joined #marpa
07:38 rns jeffreykegler: wrestling with lua test suite mostly finished (http://goo.gl/UpMHM4) so I can start testing AST serialization.
07:46 lwa joined #marpa
07:49 rns A question: I was trying to find an "official" lua formatter/beautifier to test AST serialization against, found several, but none seems to be "right".
07:50 rns Perhaps you (or anybody else) perchance can recommend something?
07:53 rns Links to what I've found are in ongoing repo file http://goo.gl/GKB7by
07:54 rns left #marpa
09:04 jeffreykegler joined #marpa
09:05 jeffreykegler rns: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9484829 -- maybe what they need is a good enough parser. :-)
09:06 jeffreykegler ronsavage: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9484434 -- thanks!
09:12 rns joined #marpa
09:14 rns jeffreykegler: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9485314 — :) well, looks like luafmt is going to be produced along the way.
09:14 rns We'll need a filter anyway and formatter seems to be a reasonable first step.
09:18 idiosyncrat joined #marpa
09:20 idiosyncrat rns: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9485381 -- That's exciting!  It'd be a really cool way to introduce ourselves to the Lua community.
09:22 rns Well, now that most of the lua test suite parses and runs ok, I planned to start seralization to formatted ast so I'll have to do it anyway.
09:23 rns Jokes aside, Lua seems to have a fair number of code analyzers/inspectors and it's so extremely simple but there must be always a job for a good parser I think. :)
09:26 idiosyncrat Back to sleep. AFK.
09:27 idiosyncrat left #marpa
09:29 rns left #marpa
09:59 rns joined #marpa
10:02 rns joined #marpa
10:02 rns metalua — http://metalua.luaforge.net/quicktour.html — looks interesting.
10:02 rns left #marpa
18:40 jdurand_ joined #marpa
18:42 jdurand_ Jeffrey - I was thinking to the most efficient way to simulate events. I supposed that using a fake terminal, and calling marpa_r_terminals_expected() to detect this is a fake temrinal could do it.
18:42 jdurand_ I also supposed that marpa_r_terminals_expected() is less expensive than marpa_r_progress_report_xxx() family.
18:43 jdurand_ And finally wondered is this is really less expensive than trigger_events(), which internally is doing a bv_scan() isn't it
19:21 jeffreykegler joined #marpa
19:23 jeffreykegler jdurand: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9488355 -- expecting fake terminals might be a cheap way of simulating events.
19:24 jeffreykegler In one of my blog posts I called this the "bigfoot technique" -- the idea being that you find it nice to believe the fake terminals are out there, but you don't actually want to come up face to face with one.
19:25 jeffreykegler I plan some experiments along those lines in fact.
19:27 jeffreykegler jdurand: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9488360 -- yes, trigger_events() does bv_scan()'s but they are cheap -- in fact they are done even when there are no events, and as you showed, the cost is minimal ...
19:29 jeffreykegler if you're used to the old way of determining efficiency, counting operations, they're surprisingly fact, because they're the kind of ops that are easy to pipeline, and they wind up almost free.
19:30 jeffreykegler * "surprisingly fact" -> "surprisingly fast"
19:31 jeffreykegler rns: re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9485705 -- metalua is, indeed, interesting. I think that Marpa and Lua may have a lot to offer each other.
19:35 jeffreykegler Jdurand: re bigfoot terminals, one thing to watch out for -- you don't want to make any right recursions ambiguous, because the will disable the Leo optimization, and they'll go quadratic.
19:36 jeffreykegler That is, beware of taking "A ::= x A" and turning it into two rules "A ::= x A  ; A ::= x A bigfoot"
19:49 flaviu1 joined #marpa
21:56 ronsavage joined #marpa
22:19 flaviu joined #marpa
22:26 ronsavage CQ: Re http://irclog.perlgeek.de/​marpa/2014-10-10#i_9484466 Yes 2.096000 is the default/standard downloadable from CPAN.
22:27 ronsavage left #marpa
22:27 ronsavage joined #marpa
22:29 ronsavage CQ: If you look closely at the line which says Jeffrey Kegler Marpa-2.096000 you'll see a down-arrow-head between Kegler and Marpa. Clicking opens a menu showing official /and dev/ releases. Choose 2.097_001 (the '_' indicates a dev version), and you could then choose to download it.
22:30 ronsavage My policy is that Marpa is almost the only package (not just for Perl) where I would ever consider installing a dev version.
23:31 jeffreykegler joined #marpa
23:32 jeffreykegler ronsavage: Re  my dev versions, thanks for the kind words!
23:32 jeffreykegler I am heavily indebted to a bold few (Ron among them) who do use my dev versions and provide test and other feedback.
23:33 jeffreykegler If you're not one of those giving me feedback, however, you should stick to the indexed, stable versions.  Because mistakes do happen.
23:57 ronsavage Also, I should add I keep at least 1 non-dev version for re-installation if a later dev version causes problems.

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