Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-12-21

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

All times shown according to UTC.

Time Nick Message
00:02 Idiosyncrat joined #marpa
00:41 Idiosyncrat My new blog post: http://jeffreykegler.github.io/Ocean-of-Awareness-blog/individual/2015/12/topdown.html
00:42 Idiosyncrat_ joined #marpa
01:42 ronsavage Good post. And it maintains your reputation as A Voice in the Wilderness :-).
02:08 Idiosyncrat_ Actually, I'm kind of tired of being a voice in the wilderness. :-)
03:57 ceridwen I'm approaching the point where I actually need to start benchmarking/testing, and I have a question: what are there in the way of *actually* context-free languages/data formats in the real world?  There are lots of toy examples that have CFGs, like say the lambda calculus, but when it comes to stuff with large corpuses that I could use, my list is kind of limited.  I think JSON is.  What else?  I've heard arguments either way for Java.
04:15 Idiosyncrat_ In the Marpa world, JSON is pretty much our standard test language.  Large corpus, and lots of parsers for benchmarking against.
04:17 Idiosyncrat_ In the Marpa world we've also used C, HTML, and Perl.
04:18 Idiosyncrat_ The web of course offers a huge corpus of (very liberal) HTML.
04:18 Idiosyncrat_ and we've used the Perl 5 distro for the C & Perl corpus -- it's as big as we needed.
04:19 VsyachePuz Idiosyncrat: this is not the article i wanted to read
04:21 VsyachePuz Idiosyncrat: it chaw the obvious thing to death, but doesn't show the way to go
04:22 VsyachePuz Idiosyncrat: It only names that way, but give no explaination of it
04:23 Idiosyncrat_ I've done quite a few posts in the past about the vision of grammar-driven programming.
04:23 Idiosyncrat_ They've consistently been my least read and most misunderstood posts.
04:24 VsyachePuz you should add links to that indepth articles at least
04:26 Idiosyncrat_ In this case the vision comes via a Perl 6 advent article.
04:27 Idiosyncrat_ Yes, I could have done more in that direction, but my rule is that when I do half a dozen posts on a topic that get misunderstood or ignored ...
04:27 Idiosyncrat_ I reprioritize my time.
04:28 VsyachePuz where I can find a detailed example of bottom-up parsing?
04:28 VsyachePuz Is there a glossagy of parting terms ?
04:28 VsyachePuz *glossary
04:29 VsyachePuz *parsing
04:29 Idiosyncrat_ The wikipedia articles on yacc/bison/LALR -- give me a sec and I'll glance at them
04:29 VsyachePuz Idiosyncrat_: the wikipedia articles are all for top down, isn't it ?
04:30 Idiosyncrat_ https://en.wikipedia.org/wiki/LR_parser
04:30 ceridwen http://blog.reverberate.org/2013/07/ll-and-lr-parsing-demystified.html
04:31 ceridwen http://blog.reverberate.org/2013/09/ll-and-lr-in-context-why-parsing-tools.html
04:32 VsyachePuz somehow that is not what I want to see
04:32 ceridwen C, (real-world) HTML, and Perl aren't context-free, are they?  I mean, you can probably construct a possibly-ambiguous grammar for them that would be context-free.
04:33 Idiosyncrat_ ceridwen: I think C is CF, HTML is.  But you're right Perl 5 is not CF.
04:34 Idiosyncrat_ ceridwen: In any case, the best starting corpus in our experience was also your first guess: JSON.
04:34 ceridwen Right.
04:35 Idiosyncrat_ VsyachePuz: are you looking for an intro to the various parsing techniques?
04:35 VsyachePuz Idiosyncrat_: no, no, no. Internet is flooded by intros
04:35 ceridwen I'm going to get to the non-context-free languages eventually :).  But that will take a lot more work, so I'm starting simple.
04:39 Idiosyncrat_ Re my post -- as discussed earlier in this channel, my intended audience was ...
04:40 Idiosyncrat_ 1.) people already interested in grammar-driven programming,
04:40 Idiosyncrat_ 2.) who plan to use a top-down approach to reach the goal.
04:41 Idiosyncrat_ Specifically, I had in mind a group of people in and around the Perl 6 community.
04:42 Idiosyncrat_ I tried to make it as meaningful as possible to as many people as possible, but its job ...
04:42 Idiosyncrat_ like that of my previous post, was to put my thoughts about top-down and LOP on record ...
04:43 Idiosyncrat_ and convey them to an expert audience.
04:44 Idiosyncrat_ I'd love to see folks start doing stuff around LOP, and of course I think Marpa is the way to do it.
04:45 Idiosyncrat_ And I'd wanted to tell those folks who are waiting for top-down to get good enough for LOP that it just ain't never gonna happen.
04:48 Idiosyncrat_ It's getting late CA time.
04:48 Idiosyncrat_ Good night!
05:34 ronsavage joined #marpa
08:39 ronsavage joined #marpa
08:41 ronsavage Actually, you position reminds me of that ancient Greek who was doomed to tell the truth about top-down parsing, but was also doomed to never be believed. Unfortunately, she's not an androgynous character, but is a female called Cassandra (https://en.wikipedia.org/wiki/Cassandra).
08:52 VsyachePuz ronsavage: warriors come to Cassandra and ask: how to organise the defense. Cassandra answers - it's not in my interest to organize the defense myself. My interest is to sell my predictions...
08:55 VsyachePuz or, "i will not help you protect your part of city, I will start from protecting mine house first"
08:58 VsyachePuz With such strategy from Cassandra, it's understandable, why warriors listen to other commanders.
09:06 VsyachePuz Other commanders says - we give you weapons and methods, which uses the technology blessed by your gods and fits to your monastery training. Might be it is slightly worse in theory, but instead it was proven in practice.
09:39 ronsavage VsyachePuz: That sounds like what Sigmund Freud called 'The Reality Principle'.
09:45 sivoais_ joined #marpa
09:50 VsyachePuz Idiosyncrat_: So, I think  Idiosyncrat should prepare tutorials for craft masters, which should build weapons for all religions on the new principle. Instead of calling warriours to his on guild for training
09:50 VsyachePuz *to his own guild
09:52 VsyachePuz If he wish to extract the value from this, he should establish agreements with craftmasters for royalty paymants
09:55 sivoais joined #marpa
10:05 sivoais joined #marpa
10:15 sivoais joined #marpa
10:25 sivoais joined #marpa
10:35 sivoais joined #marpa
10:45 sivoais joined #marpa
10:55 sivoais joined #marpa
11:05 sivoais joined #marpa
11:15 sivoais joined #marpa
11:25 sivoais joined #marpa
11:35 sivoais joined #marpa
11:45 sivoais joined #marpa
11:55 sivoais joined #marpa
12:05 sivoais joined #marpa
12:15 sivoais joined #marpa
15:09 sivoais joined #marpa
15:21 sivoais joined #marpa
15:31 sivoais joined #marpa
15:41 sivoais joined #marpa
15:51 sivoais joined #marpa
15:59 Idiosyncrat joined #marpa
16:01 sivoais joined #marpa
16:12 sivoais joined #marpa
16:21 Idiosyncrat_ ronsavage: re http://irclog.perlgeek.de/marpa/2015-12-21#i_11750649
16:22 sivoais joined #marpa
16:22 Idiosyncrat_ The difference between me & Cassandra is that she was bringing bad news, and my news is good.
16:23 Idiosyncrat_ But I have thought that I should claim Warnock's title. :-)
16:32 sivoais joined #marpa
16:42 sivoais joined #marpa
16:44 koo8 joined #marpa
16:52 sivoais joined #marpa
16:57 sivoais joined #marpa
19:59 Idiosyncrat_ There will probably be a third in my current series of blog posts on top-down and LOP
20:00 Idiosyncrat_ The last one was about the basic limits of top-down --
20:01 Idiosyncrat_ the first one had been about top-down's unpredictability and its effect on grammar reuse, which is essential for LOP.
20:02 Idiosyncrat_ The next one will probably be about grammar reuse from the point of view of speed ...
20:03 Idiosyncrat_ That is, for LOP the parsing technology must be powerful enough for the grammars on interest ...
20:03 Idiosyncrat_ it must allow you to combine two of them and be able to predict the result ...
20:03 Idiosyncrat_ and, if the two pieces run in linear time, the result should run in linear time ...
20:04 Idiosyncrat_ or at least it should do so most of the time, and the cases where it might not should be predictable (as is the case with Marpa).
21:08 ronsavage joined #marpa
21:11 ronsavage JK: Re: http://irclog.perlgeek.de/marpa/2015-12-21#i_11753274. As appropriate as Warnock is (https://en.wikipedia.org/wiki/Warnock%27s_dilemma) I'm reluctant to make it Marpa's motto!

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