Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-10-17

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

All times shown according to UTC.

Time Nick Message
00:55 idiosyncrat_ joined #marpa
01:48 ilbot3 joined #marpa
01:48 Topic for #marpa is now Start here: http://savage.net.au/Marpa.html - Pastebin: http://scsys.co.uk:8002/marpa - Jeffrey's Marpa site: http://jeffreykegler.github.io/Marpa-web-site/ - IRC log: http://irclog.perlgeek.de/marpa/today
03:37 idiosyncrat_ joined #marpa
03:57 idiosyncrat_ Good night!
06:28 ronsavage joined #marpa
07:36 sirdancealot joined #marpa
09:26 sirdancealot joined #marpa
13:30 sirdancealot joined #marpa
16:07 Cheery While ago Keggler was about to upgrade his paper. It's been a while. Did you finish your new version of your marpa paper yet?
17:13 idiosyncrat_ joined #marpa
17:13 idiosyncrat_ Cheery: re https://irclog.perlgeek.de/​marpa/2016-10-17#i_13417850
17:14 Cheery idiosyncrat_: listening. :)
17:14 idiosyncrat_ The new, expanded paper turned out to be much more work than I anticipated.  I shelved it for now.  Sorry.
17:14 Cheery okay.
17:15 Cheery I should still improve my parser some weekend. And I think I should point out stronger where the origins are.
17:15 idiosyncrat_ I try to estimate effectiveness in terms of audience / amount-of-work, and thought the paper upgrade would be a couple of weeks.  I severely misestimated.
17:16 idiosyncrat_ So I'm working on Marpa::R3/Kollos -- they also are taking a lot of time but the audience is much larger.
17:16 Cheery idiosyncrat_: it's ok. I tend to diverge from doing things I meant to do at times too.
17:17 Cheery Also, the parser I wrote based on marpa has been an amazing asset when designing my language.
17:17 idiosyncrat_ Cheery: I'm glad to hear that.
17:17 Cheery I already wrote a C parser with it.
17:18 idiosyncrat_ By the way, for those listening in, funding can alter my priorities.
17:18 Cheery https://github.com/cheery/​lever/blob/master/lib/c.lc
17:18 Cheery https://github.com/cheery/lever​/blob/master/lib/ansi-c.grammar
17:19 idiosyncrat_ Right now, I'm self-funded for all of it, so I do it based on estimates of audience / amount-of-work
17:19 idiosyncrat_ Cheery: I'm happy to hear my work was helpful!
17:20 Cheery idiosyncrat_: got a link where to fund you?
17:20 Cheery or a collection of ways.
17:21 idiosyncrat_ Right now I use Paypal -- it's my jeffreykegler at jeffreykegler dot com mailing address.
17:21 idiosyncrat_ If it's a serious amount of money, send do a small test first.
17:24 Cheery I'd be probably able to send 100$ at max. But I think this is one of few things I would want to recompensate.
17:26 Cheery I will still try to figure out the right recursion optimization though.
17:27 Cheery but these are unrelated. I won't send you money because you would help me now. I would send it because you already helped me out a lot.
17:30 Cheery idiosyncrat_: I remember I tried to "rediscover" the right-recursion optimization few times. It failed because it's not simple.
17:31 Cheery What it should do is obvious.
17:31 Cheery But what else does the particular implementation cause is not.
17:31 Cheery Yeah. Validity.
17:31 Cheery I think I should approach it from that angle.
17:32 idiosyncrat_ I couldn't change priorities for $100. :-)
17:32 Cheery I know. :)
17:33 idiosyncrat_ But any gesture is appreciated.
17:33 idiosyncrat_ One thing I wanted to do in the paper is re-explain Joop's right-recursion optimization so that it is more intuitive.
17:34 idiosyncrat_ I came up with something I thought nice, but I got to the point where I just didn't think I could justify the time.\
17:34 Cheery Would it be something I am able to complete?
17:35 Cheery Basically I understand the algorithm really far here.
17:35 Cheery But didn't figure out the right-recursion rule.
17:36 Cheery or well.
17:36 Cheery I understood about this:
17:37 Cheery When this optimization has effect is a situation that you have a recursively growing solution.
17:38 Cheery basically same symbol matches recursively, forming a sequence from origin.
17:38 Cheery such as 1, 3, 5, 8 etc.
17:39 Cheery it causes the chart to grow into sort of a triangle.
17:39 Cheery if you plot the chart counts.
17:39 Cheery the optimization records only one result, and derives the others from it.
17:39 Cheery or the two extremes.
17:40 Cheery There was that you had some way to recognize the right-recursive symbols, to optimize it further.
17:40 Cheery So you wouldn't need to do the checking on symbols that cannot right-recurse.
17:44 idiosyncrat_ Cheery: https://github.com/jeffreykegler/Marpa-theory.git
17:45 idiosyncrat_ That's the repo -- it contains the paper as I left it.  (It's now a book actually, and will be The Marpa Book when it comes out.)
17:46 idiosyncrat_ But I really can't imagine anyone else helping me carry it forward at this point.  Alas.
17:51 Cheery I think I'll try to figure this out. Then I will write a blog post. That way there's one explanation more for this concept.
17:52 Cheery and one more to help you source it into the book.
17:52 Cheery It probably doesn't help that I write few lines there. :)
17:52 Cheery or does it?
17:53 idiosyncrat_ It won't hurt to try.
17:54 Cheery It's up to you. Which one would you prefer? A blog post or direct contrib?
17:54 idiosyncrat_ Blog post.
17:55 idiosyncrat_ Please credit the early draft of the The Marpa Book.
17:55 idiosyncrat_ IIRC in the Marpa book I revised repeatedly and each revision presented the ideas in a way I liked better.  That's the good news.
17:55 Cheery Will do that. I'm in middle of installing dependencies to compile a pdf to read.
17:56 idiosyncrat_ Bad news is that I didn't completely rewrite after each presentation, so the current draft contains several different & not completely consistent approaches.
17:57 idiosyncrat_ Along with a lot of repetition.
17:57 idiosyncrat_ It's a mess.
17:59 Cheery repetition is not necessarily bad thing.
17:59 idiosyncrat_ joined #marpa
17:59 Cheery repetition is not necessarily bad thing.
17:59 idiosyncrat_ Not necessarily, but in this case I predict you'll get sick of it. :-)
18:00 Cheery I'll remember to mark down when I get sick of it. ;)
18:02 Cheery yesterday I read about GCC internals btw.
18:02 Cheery specifically about GIMPLE and RTL
18:05 Cheery mm.. You'll probably be interested about what I've been doing with marpa algorithm, right?
18:05 Cheery https://gist.github.com/cheery/329a2c769618f6​fb317452b0f1a0d691#file-webassembly-12-schema
18:05 Cheery this is probably the first not that formal use of the algorithm.
18:05 Cheery casual.
18:06 Cheery I wanted to write the schema of webassembly in language that isn't ugly.
18:06 Cheery as far as I could.
18:06 Cheery then I translated it into easier-to-parse json
18:06 Cheery this is pretty much the usage pattern I seem to get with marpa.
18:07 Cheery lets take a complex thing, problem hard to represent with tools I have
18:07 Cheery write a completely custom language to tackle it.
18:07 Cheery use a standard tokenizer
18:08 Cheery then produce a mid-result that is fast to load.
18:08 Cheery and later use that to drive something.
18:10 Cheery Then upwards-flowing-documentation and I got a really maintainable program.
18:14 Cheery There's little bit of trouble in the approach though.
18:15 Cheery It pushes me to make designs that are very much clockwork like.
18:16 Cheery with abstractions meant to simplify or remove internal details erased.
18:17 Cheery It's probably not ever going to be a popular way to design a language. :)
18:22 idiosyncrat_ It's interesting looking.
18:43 idiosyncrat_ left #marpa
18:43 idiosyncrat_ joined #marpa
19:29 Cheery joined #marpa
19:50 sirdancealot joined #marpa
20:19 sirdancealot joined #marpa
21:39 ronsavage joined #marpa
23:37 idiosyncrat_ joined #marpa

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