# IRC log for #marpa, 2015-11-12

All times shown according to UTC.

Time Nick Message
00:04 Idiosyncrat joined #marpa
01:25 Idiosyncrat joined #marpa
01:46 Idiosyncrat As many of you will know, Marpa is the 1st implementation of Leo's algorithm.
01:46 Idiosyncrat And I've been a little surprised that so far I've had no imitators.
01:47 Idiosyncrat s/1st/1st practical/
01:48 Idiosyncrat Joop did a small experimental implementation apparently around 1990 -- the focus seems to have been assuring himself that the theory was sound ...
01:48 Idiosyncrat and not on creating anything useable in practice.
01:49 Idiosyncrat Now that I'm re-stating Leo's method from 1st principles, I think I see why I've (so far) had no imitators.
01:50 Idiosyncrat It is difficult, and to describe how to actually use it in practice requires making a lot of decisions Joop left open.
01:51 Idiosyncrat I've already solved all these issues in the C code, and there I did proofs as necessary.
01:51 Idiosyncrat Now I'm "translating" C into math, with proofs and sound definitions for everything.
01:53 Idiosyncrat I think at this point many more people come to know Joop's method from my current theory paper, than from Joop's paper.
01:54 Idiosyncrat Possibly this new paper will take that one step further, and we will start to see implementations like lotsawa incorporate Leo's method.
01:55 Idiosyncrat and tutorials which cover Leo's method the way several good ones cover Earley's algorithm today.
03:26 ronsavage joined #marpa
03:27 ronsavage I've re-worded a couple of places a the start of my new article, to clarify some background in Marpa is expected.
04:23 Idiosyncrat joined #marpa
04:23 Idiosyncrat ronsavage: Starting a 2nd more careful reading.
04:24 Idiosyncrat A Perl style question, not necessarily a suggestion:
04:24 Idiosyncrat Stuff like \$\$t2{phrase} = "(\$\$t1[2][0])";
04:25 Idiosyncrat I read somewhere (Damian?) that you should clarify multiple sigils with curlys.
04:26 Idiosyncrat So I think that LHS is equivalent to \${\$t2{phrase}} and Damian(?) says that is clearer.
04:30 Idiosyncrat 2nd observation: on a hasty reading, I wondered, "where is the input Ron is using?"
04:30 Idiosyncrat It's there, but it's not in display format.  Maybe if it were, it would be harder to miss and more "orthogonal"
04:35 Idiosyncrat Finally, I'll note that if and when you want flatter data, the ":first" action comes in handy -- in cases where an array of a single element is being returned ...
04:36 Idiosyncrat you get instead the single element.
04:37 Idiosyncrat That's just sort of an aside.  The example data in the article is not overly deep, and if you flatten it, it may not be as good an example.
04:37 Idiosyncrat That's all on this 2nd reading.
04:37 Idiosyncrat Thanks!
04:46 ronsavage JK: I'll go thru the article again, but briefly: \$\$t2{phrase} = "(\$\$t1[2][0])"; is my preferred way. Other people prefer as you suggest. I would rather the article use my real code.
04:47 ronsavage And the input, yes, I'd better emphasize that.
04:53 Idiosyncrat joined #marpa
05:23 ronsavage Also, how do I use POD to generate 'display format'?
05:24 ronsavage I'm trying C<...> and it looks better already.
13:24 shadowpaste joined #marpa
16:07 djns joined #marpa
17:51 Idiosyncrat joined #marpa
17:54 Idiosyncrat ronsavage: I was using "display format" in a loose sense.  You present your input "inline", rather than breaking it out.  You have code in "display", so whatever means you use to do that, could display the input.
17:54 Idiosyncrat Re "\$\$t2" -- that's fine, as long as your choice of that style is considered and deliberate.
18:54 ilbot3 joined #marpa
18:54 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
20:35 ilbot3 joined #marpa
20:35 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
21:20 ronsavage joined #marpa
21:20 ronsavage Re "\$\$t2". Yeah. Some people use \$t2->{...} but I reserve -> for method calls.
23:55 Idiosyncrat joined #marpa
23:56 hobbs joined #marpa
23:56 hobbs joined #marpa