Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-06-24

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

All times shown according to UTC.

Time Nick Message
12:18 jeffreykegler joined #marpa
15:51 jeffreykegler joined #marpa
15:51 jeffreykegler I think Andrew's talk just finished.  I'm not sure because my Internet is flaking out.
15:52 jeffreykegler I didn't get the sound, but the slides looked good!
15:53 jeffreykegler https://www.youtube.com/watch?v=XaScLywH2CI
15:55 Aria Ooh. I'll have to watch that when I have headphones
16:02 jeffreykegler1 joined #marpa
16:05 jeffreykegler joined #marpa
17:58 hobbs I feel like I could have done a lot better with more prep time. I ran under time, so I could have dug into Marpa features a little more
17:58 hobbs but I got a positive reaction, and Larry chatted me up afterwards
17:58 hobbs so, I didn't blow it :)
18:05 Aria Oh nice!
18:07 jeffreykegler joined #marpa
18:07 jeffreykegler I did get a auto-transcript of Andrew's talk, which seemed great.
18:08 jeffreykegler I still can't get sound -- I'll experiment tonight.
18:09 jeffreykegler Did Larry attend?  Or just seek you out afterwards?
18:09 hobbs he did attend
18:09 hobbs and interjected a little
18:13 jeffreykegler I've found Marpa hard to present myself just because I think a lot about parsing theory, the way lots of people did in the 70's and not just in academia.
18:13 jeffreykegler I worked at a time-sharing company which had a database and they found a way to work a self-compiling language into their product, just because it was the cool thing to do.
18:14 lucs jeffreykegler: May I ask where you got an auto-transcript (and do other YAPC talks have them?)?
18:14 jeffreykegler And the guys at this company all wore suits, if you can believe that, white shirt, tie, the Victor Brett/IBM look
18:14 hobbs mmhmm :)
18:14 jeffreykegler The auto-transcript is on the Youtube page
18:15 lucs Oh, I'll look more closely :) Thanks.
18:15 jeffreykegler There an icon labeled "Transcript"
18:16 lucs Aha, got it.
18:16 jeffreykegler But anyway, I wasn't even allowed into the parsing stuff because I wasn't senior enough, but I kept thinking about parsing ...
18:17 jeffreykegler And by the time I finally am ready to do something, it seems like everybody forget about it, and I'm speaking a totally new jargon.
18:18 hobbs yeah, from what I know it does seem like "all the good work" was done back then, and then there was a lull
18:18 hobbs which is why we're all using a lot of 70s stuff
18:19 hobbs I don't have the academics, just the practical end, so I'm in an interesting position. Hope I didn't misrepresent anything too badly :)
18:19 jeffreykegler So anyway, it seems like a nice simple presentation tailored to its audience.
18:20 jeffreykegler No, no mistakes (that I noticed anyway)
18:21 jeffreykegler By the way, one way to describe the class of grammars Marpa can parse is "anything you can write in BNF"
18:21 jeffreykegler Which is a very handy feature, because it allows auto-generation of grammars.
18:21 hobbs yeah, it slipped my mind, and my speaker notes dropped out on me because technology
18:22 jeffreykegler You can't do that for the input to yacc/bison, for example, because you have no way of knowing whether yacc will handle the auto-generated grammar.
18:23 hobbs that is interesting
18:23 jeffreykegler And with a little care, you can be sure the grammar will parse in linear time.  Just keep ambiguity limited, and avoid unmarked middle recursions.
18:24 hobbs define "marked" for me?
18:24 jeffreykegler Yes (re the auto-generated grammars), if you think about it, there's a lot that could be done there.
18:25 jeffreykegler Actually, the exact class of grammars that Marpa (leo 1991) parses in linear time is undecidable.
18:25 jeffreykegler So "marked" will never have a precise definition.
18:25 jeffreykegler But basically if the human eye can spot it in reading, it's marked.
18:25 hobbs fair enough
18:26 jeffreykegler For similar reasons, I used the vague word "limited" in describing the ambiguity you can get away with and stay linear.
18:27 hobbs right
18:27 jeffreykegler If you grammar stays linear, your ambiguity was "limited".  If it doesn't, it wasn't. :-)
18:27 hobbs :)
18:27 jeffreykegler Actually, my paper will contain some more careful definitions.
18:28 jeffreykegler But for the practical programmer, these rules of thumb will be more useful.
18:28 hobbs time for some YAPC mingling. Thanks for the chat, I'll be around sometime in the future :)
18:28 jeffreykegler And I'm off to lunch.
18:28 jeffreykegler hobbs: Thanks!
20:39 jdurand_ joined #marpa
20:39 * jdurand_ is looking at https://www.youtube.com/watch?v=XaScLywH2CI
21:02 jdurand_ Interesting talk - many thanks Andrew for your work and time on this
21:42 hobbs joined #marpa
21:42 hobbs joined #marpa
21:56 sivoais joined #marpa
22:12 jeffreykegler joined #marpa

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