Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-02-26

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

All times shown according to UTC.

Time Nick Message
00:23 jeffreykegler joined #marpa
03:30 atweiden joined #marpa
03:52 atweiden has anyone explored wrapping libmarpa from Nim (nim-lang.org)?
03:54 jeffreykegler Not to my knowledge
03:56 ronsavage joined #marpa
04:13 atweiden "Third, Kollos will make the higher layers of Marpa::R2, which are now tightly tied to Perl, available as a C language api"
04:13 atweiden would it be better to wait for kollos?
04:14 jeffreykegler That could be called either way.
04:14 jeffreykegler If you're content with the Libmarpa API, there is no real reason to wait.
04:14 jeffreykegler But the Libmarpa API is quite low-level -- symbols don't have names, for example.
04:14 jeffreykegler In fact, there are no strings at all, everything is an integer.
04:15 jeffreykegler Which you could view as giving you *a lot* of flexibility in representing symbol names and other strings.
04:15 jeffreykegler Or you could view as a real hassle.
04:29 atweiden well, i have a pressing use case for marpa
04:29 atweiden Marpa::R2 will work, and might even be overkill
04:29 atweiden but nim is a beautiful language with intuitive FFI, compiles down to C
04:30 atweiden i am considering putting real money on nim; if not Marpa::R2
04:31 jeffreykegler Kollos will solve the problem has having the higher-level stuff in lightweight, efficient, embeddable form.
04:31 jeffreykegler But it is *not* happening quickly, unfortunately.
04:31 jeffreykegler s/has having/of having/
04:52 atweiden in my experience funding electrum development (http://i.imgur.com/vQ1qVWd.png), typically money helps open source proceed at a quicker pace
04:52 atweiden but it does come at a price
04:53 jeffreykegler I'd love to see funding for Marpa, of course.
04:54 jeffreykegler And of course, often the funders want to see specific things done to the open source project in return for the funding.
04:55 jeffreykegler A lot of Luajit development was funded IIRC by people who wanted it to target specific platforms
05:01 atweiden yes, and bindings are in a similar spirit. i wouldn't expect bindings to necessitate low level changes to libmarpa
05:02 atweiden my use case is parsing double entry accounting logs, e.g. https://github.com/mafm/ledger.py. Marpa::R2 will do it for sure, but Nim could be more fun
05:04 jeffreykegler An interesting looking problem -- is there a grammar and/or other precise specification?
05:06 atweiden there isn't any yet. i don't know if the original ledger (ledger-cli.org) has one, but in any case the syntax i am using in my logs differs slightly, although it is most similar to @mafm's ledger.py
05:13 jeffreykegler The format is carefully described except for whitespace interpretation, but whitespace interpretion is basic to it.
05:14 jeffreykegler The general idea for the whitespace is obvious from the examples, but matters like how liberal to be seem to be left to the programmer.
05:15 jeffreykegler It's late California time -- AFK
05:49 rns joined #marpa
05:55 rns atweiden: libmarpa binding is easy using libffi -- I've been experimenting with lua and python -- https://github.com/rns/libmarpa-bindings
06:03 atweiden rns: looks like the perfect starting point
06:14 rns atweiden: Glad you liked it. BTW, I vaguely remember flaviu, who is on this channel, works with the nim lang.
06:57 rns left #marpa
08:03 basiliscos joined #marpa
10:59 lwa joined #marpa
15:29 koo7 joined #marpa
17:08 jeffreykegler joined #marpa
17:10 jeffreykegler atweiden: re http://irclog.perlgeek.de/m​arpa/2015-02-26#i_10181925
17:12 jeffreykegler you might consider prototyping the parser via Marpa::R2, to nail down issues of how liberal/conservative to be, whether to use events or try to go entirely syntax-driven, etc.
17:12 jeffreykegler Once you know what you want to implement, you can translate it into the nim version.
18:28 jeffreykegler I've been converting my public resume into HTML form, and in the process updating it --
18:28 jeffreykegler http://jeffreykegler.github.io/Curriculum-vitae/
18:29 jeffreykegler I suspect HTML is more practical for a resume these days.
18:30 jeffreykegler It's an unusual resume, of course, because the only thing on it for the past 7 years is an open source project -- Marpa.
18:48 rns joined #marpa
18:55 rns left #marpa
20:05 koo7 who wrote this shitty code? https://github.com/rns/libmarpa-binding​s/blob/master/python/libmarpa_codes.py
20:25 basiliscos joined #marpa
20:53 ronsavage joined #marpa
22:19 flaviu koo7: What's wrong with it?
22:21 koo7 flaviu, on a second thought, not much, given that checking the numbers is a good idea and making the least modifications from the C original, i probably couldnt have done it much better
22:25 rns joined #marpa
22:27 rns left #marpa
22:35 ernimril joined #marpa
22:53 lucs koo7: Why do you call it shitty code?
23:02 koo7 its mine
23:02 lucs Aha! :)
23:02 koo7 :)
23:03 lucs It came off sounding not very nice for rns though :/
23:03 koo7 he knows its mine:D
23:04 lucs Yeah, but it confuses the lurkers.
23:04 lucs Like me.
23:05 koo7 ok, i will try to only speak in praises about my code from now on here
23:11 * koo7 is sometimes a troll and cant help it
23:21 ronsavage R U a troll or auto-troll?
23:23 koo7 whats an auto-troll?

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