Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-01-16

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

All times shown according to UTC.

Time Nick Message
00:08 sivoais joined #marpa
01:00 jeffreykegler http://blogs.perl.org/users/jeffrey_kegler/2010/06/jay-earleys-idea.html
01:01 jeffreykegler Ooops! make that http://jeffreykegler.github.io/Ocean-of-Awareness-blog/individual/2010/06/jay-earleys-idea.html
01:02 jeffreykegler http://en.wikipedia.org/wiki/Earley_parser
01:03 jeffreykegler In connection with the FLOSS show, a few folks have asked how to get a conceptual fix on what Marpa can do -- how Matpa "thinks" -- what makes it tick.
01:04 jeffreykegler If you think of Marpa as an Earley's parser with other stuff, you've basically "got it"
01:05 jeffreykegler The above two links are (a blog post of mine and the Wikipedia article) are two of many places which describe Jay Earley's basic idea.
01:05 jeffreykegler Read them, think about it, compare with what you know of other parsing approaches, speculate about how you'd improve it ...
01:06 jeffreykegler and once you've done so, you'll have developed the basic intuitions necessary to understand Marpa.
01:09 jeffreykegler There is also this one: http://loup-vaillant.fr/tutorials/earley-parsing/recogniser
01:12 lucs joined #marpa
01:13 jeffreykegler Marpa/Leo does a lot of stuff the original Earley's algorithm cannot do, but 90% plus of what's involved in thinking about Marpa is thinking in terms of Jay Earley's original sets and items.
02:06 flaviu joined #marpa
02:20 ronsavage joined #marpa
03:21 jeffreykegler While I'm at it, re thinking about Earley sets/items, another suggestion --
03:22 jeffreykegler In first trying to understand it, think of it as a data structure and not an algorithm.
03:23 jeffreykegler That is, think of a parse as described by Earley items in Earley sets, assuming that there's some efficient way to come up with the Earley items/sets.
03:24 jeffreykegler A lot of accounts teach the data structure via the algorithm, and in my experience that's getting it backward.
04:05 Guest87194 Totally.
05:29 ronsavage joined #marpa
05:47 jeffreykegler rns: In case you wonder why I'm fooling around in the Kollos repository -- I had to refactor Libmarpa due to the "discard events" change, which was an unforeseen complication, but something that had to be done sooner or later anyway.
05:48 jeffreykegler After the refactoring, I wanted to make sure that Libmarpa actually builds somewhere else than for Marpa::R2, so I'm rebuilding Kollos with the new Libmarpa ...
05:48 jeffreykegler which ran into various packaging glitches now fixed.
05:48 jeffreykegler rns: I think you're doing something like Kollos in parellel, so I thought I'd keep you informed.
05:52 rns jeffreykegler: Thanks for explaining. I saw packaging activity in Kollos repo and your info came just in time.
05:53 rns What I'm doing in libmarpa-bindings is basically a lua binding for libmarpa.
05:55 jeffreykegler Yes, I assume you're trying to keep it possible to merge our 2 efforts, so I thought an explanation of my activity was in order.
05:55 jeffreykegler In the meantime, I hope for interesting news from your effort.
05:57 rns Me too. :)
05:59 rns Re merging: eventually, I'm planning to write an extensible lua parser in lua as a libmarpa application, sort of 'metalua done right', which will be able to parser LUIF.
05:59 jeffreykegler The Lua folks should find that exciting, I'd think.
06:00 jeffreykegler I did remember thinking, while reading the MetaLua docs, that those guys could have really created a very cool tool, given a halfway decent parser. :-)
06:01 rns :) Yes. BTW, metalua saw an update recently.
06:01 rns I was thinking about smart lua preprocessor based on that extensible parser.
06:02 rns A preprocessor that would allow syntactyc extensions, e.g. assignment to multiple keys in table constructors.
06:02 jeffreykegler I'll be very much looking forward to it.  Did you get a chance to look at the FLOSS segment?
06:03 rns Yep.
06:03 rns I've downloaded the HD version.
06:03 rns Randall Swartz and Andrew did very well, from what I've seen.
06:04 rns And, backlogging your conversation with Randall on this channel, I recalled that Steve Yegge once said perl is so successful because Larry is a marketing genius.
06:05 rns It was nice to seel in person the driving force behind that -- Randall Swartz.
06:06 jeffreykegler Larry had a serious rep when Perl came out, and has a good public persona, but I'm not sure I'm call him a marketing genius.
06:07 jeffreykegler The marekting was really catalyzed & driven by Randal.
06:07 rns Well, Yegge was trying to explain the success of perl speaking about languages in general.
06:07 rns Overall, I'm not a big fan of marketing, but that FLOSS show made me changing my mind -- very useful presentation of a very cool technology.
06:08 jeffreykegler Yegge is insightful, but I think he called this one wrong -- though I've never seen the quote you refer to.
06:08 jeffreykegler As Randal said on this channel, Perl might well have succeeded without him.
06:10 jeffreykegler Larry was already known for the patch tool, and rn, which was the most popular newsreader at a time when news was a very big medium for us geeks.
06:11 jeffreykegler But I do believe Perl *was* the first Internet tool to have something like marketing done for it, so Yegge is on the mark in that sense.
06:12 jeffreykegler Before that, we all expected good software to succeed on its merits -- we were really stupid in those days. :-)
06:12 rns It was "Larry Wall is a formidable marketeer, better at it than most Harvard business school graduates." so I misquoted a bit.
06:13 jeffreykegler rns: you definitely caught the sense, and in this case I cannot say that I agree with Yegge.
06:14 jeffreykegler It's more like Randal and Tim O'Reilly, two genuine marketing genuises, recognized Larry's talent, and thrust him into the role ...
06:14 jeffreykegler which, to be sure, he does fill quite well.
06:15 jeffreykegler I've known both of them from the mid 90's, by the way.
06:16 jeffreykegler My history in the Perl community is kind of wierd .. I initially knew everybody, because Larry and Randal were all the people there was to know ...
06:16 rns :)
06:17 jeffreykegler but the community grew, and I went underground, and by the time Marpa comes out, I don't know most folks and they have no idea who I am.
06:19 rns I see.
06:21 jeffreykegler Good night.  rns: your MetaLua++ effort is very interesting.  Thanks!
06:21 rns Good night.
06:50 rns_ joined #marpa
06:51 koo5 joined #marpa
06:51 rns left #marpa
06:52 rns_ left #marpa
07:51 basiliscos joined #marpa
08:21 basiliscos joined #marpa
12:39 AbigailBuccaneer joined #marpa
15:18 lwa joined #marpa
15:56 jeffreykegler joined #marpa
17:30 koo5 joined #marpa
17:43 basiliscos1 joined #marpa
18:10 basiliscos joined #marpa
19:35 jdurand joined #marpa
19:35 jdurand Anybody experiencing failure with their Marpa perl package /only/ on openbsd ? C.f. http://www.cpantesters.org/cpan/report/a3087c0e-9a76-11e4-b221-9e126cbd7f71 - this is puzzling me
20:16 jeffreykegler joined #marpa
20:17 jdurand joined #marpa
20:17 jdurand Forget previous link & apologizes - I meant: http://matrix.cpantesters.org/?dist=MarpaX-Languages-ECMAScript-AST%200.020
21:26 ronsavage joined #marpa
21:49 sivoais joined #marpa
22:03 koo5 joined #marpa
22:42 flaviu joined #marpa
22:51 sivoais joined #marpa

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