Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-12-16

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

All times shown according to UTC.

Time Nick Message
00:48 ronsavage I've updated Marpa's homepage with many modules found on CPAN and github.
01:34 jeffreykegler joined #marpa
01:34 jeffreykegler flaviu: re the "friend" layer
01:35 jeffreykegler I assume you're reading the cweave output -- and I'm glad people are looking at it ...
01:36 jeffreykegler but there's material in there from old design thoughts, that has never been updated.
01:36 jeffreykegler I apologize because it's a hard enough document to read without that kind of issue, but I've never had time to revise it.
01:38 jeffreykegler In the way of excuses, Marpa is a project of staggering size for (what was until recently) one person, and I've been forced to be ruthless about doing what was most likely to
01:38 jeffreykegler 1.) be responsive to actual users; and
01:38 jeffreykegler 2.) attract new ones.
01:39 jeffreykegler I'll look at marpa.pdf and possibly revise it, since at this point it has at least 2 readers besides myself, and thanks for pointing out this issue.
01:39 flaviu jeffreykegler: Don't worry about it, I'm not really reading it - just scanning over it a bit.
01:40 jeffreykegler Scanning counts as reading for this purpose ...
01:40 jeffreykegler With that kind of material I always start with a skim-over.
01:41 jeffreykegler The fixes should be quick -- with old design material, I will probably just delete it.
01:41 jeffreykegler And I find it's good, when possible, to respond to these things while they're fresh in the mind.
01:43 jeffreykegler ronsavage: many, many thanks for doing this -- there sure is a bunch of stuff out there at this point.
01:45 flaviu Re. attracting new users: Perhaps a nice modern-looking website would be good. Frankly, the current pages look dated, although that is very excusable given that the website is not the major goal of the project.
01:45 flaviu I'd like to nominate myself - I have winter break coming up without a project to work on.
01:46 jeffreykegler flaviu: that'd be great!
01:48 jeffreykegler ronsavage: what do you think -- a website that concentrates more on it's look, so we can focus more on technical, reference and archival issues?
01:50 jeffreykegler flaviu: by the way, I'm hacking up the CWeb file now -- I've deleted that whole section on layers.
02:51 jeffreykegler I just uploaded a fixed version of marpa.pdf -- the Cweb output of Marpa's source code: https://drive.google.com/file/d/0B9_mR_M2zOc4WjI1dU02QnhVQVU/view?usp=sharing
02:52 jeffreykegler It has a fix suggested by flaviu plus some added comments on the code for finding the nullable symbols in a grammar.
03:09 ronsavage joined #marpa
03:11 ronsavage Is anybody re-writing Marpa::HTML?
04:37 ronsavage Ignore my last question. I was confusing Marpa::HTML with Marpa::R2::HTML.
06:22 jeffreykegler joined #marpa
06:23 jeffreykegler I wrote up a note on Loup's "empty rules" tutorial, suggesting another algorithm from the Libmarpa code: https://github.com/jeffreykegler/kollos/blob/master/notes/misc/loup2.md
06:24 Aria Oh excellent.
06:30 jeffreykegler Aria: yes, this algorithm is one of the two "work horses" of the grammar precomputation.
06:30 jeffreykegler The other is the implementation of Warshall's algorithm.
09:44 ronsavage joined #marpa
10:00 ronsavage I have no problem with a website dedicated to Marpa. marpa.(com|net|org) have been taken. We were asleep at the wheel, there :-( Do you want me to register marpa.com.au and/or marpa.net.au?
10:03 lwa joined #marpa
10:44 jeffreykegler joined #marpa
10:45 jeffreykegler ronsavage: our chances of beating out the Mid-Anglia Rail Passengers Association were never good. :-)
10:50 jeffreykegler flaviu: if you want one or both of the two .au names, let ronsavage know.
11:22 ronsavage Actually, one site said marpa.org has expired, but another said it hadn't. I will include it in any order I submit, and it'll be on backorder.
11:33 flaviu joined #marpa
13:49 slothmachine2 joined #marpa
18:09 lwa joined #marpa
18:14 jeffreykegler joined #marpa
18:23 jdurand__ joined #marpa
18:23 jdurand__ ronsavage: Re http://irclog.perlgeek.de/marpa/2014-12-15#i_9806923 - thx -;
18:34 sirdancealot joined #marpa
18:40 jeffreykegler To my note on Loup's "empty rules" post, I've added a proof that my suggested alternative algorithm is linear time in the size of the grammar: https://github.com/jeffreykegler/kollos/blob/master/notes/misc/loup2.md
18:44 jeffreykegler Loup's approach is worst case quadratic -- this can make a difference for big grammars, and also for "2nd order language" techniques, where you generate mutliple grammars on the fly.
19:06 jeffreykegler Actually, I realized my complexity analysis "cheated", because the examining the RHS of each rule introduces an additional factor of the size of the grammar.
19:07 jeffreykegler So Loup's approach is actually cubic O(n**3) and my suggested alternative is actually quadratic O(n**2), where n is the symbol count of the grammar.
19:09 jeffreykegler This actually makes a stronger case for the alternative, because quadratic should be do-able for large, but realistically sized, grammars.  Cubic catches up with you a lot faster, and is more likely to make a difference even in practical cases.
20:38 flaviu joined #marpa
20:41 jdurand joined #marpa
20:42 jdurand rns: I +1'ed at https://stackoverflow.com/questions/27438624/how-to-extract-function-definition-from-c-source-file despite it is on-hold -;
20:48 jeffreykegler I understand the motivation behind the hold policy, which is to ban questions like, "What is the best programming language to use?"
20:53 jeffreykegler But it seems to be being applied to some perfectly reasonable and interesting questions.  As a hypothetical example, if I want to I ask "is there a good package for doing Fourier analysis in C?", am I really supposed to try out a Fourier analysis, fail, and report my attempt in order to pose the question?
20:56 flaviu reading the logs, I think that it might be best to wait for the domain name. I can use github pages for now, and I don't actually have anything written at the moment.
20:57 jeffreykegler flaviu: github is good, because a lot of us are used to it as a way to report issues, and submit changes.
20:58 jeffreykegler flaviu: just to be clear, I take it you'll be focusing more on issues of appearance -- on coming up with a site that looks attractive to modern eyes?
20:58 flaviu Yes. I don't understand much of how marpa works at the moment
20:59 jeffreykegler flaviu: so you'll be more concerned with presentation than content, then.
21:02 jeffreykegler That's be good, because apparently a lot of folks do want a certain "look" in a website, and when they don't see might judge the whole project is somehow "not real"
21:53 ronsavage OK. Domain name stuff on hold.
22:12 LLamaRider joined #marpa
22:54 sirdancealot joined #marpa

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