Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-11-20

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

All times shown according to UTC.

Time Nick Message
00:21 ronsavage left #marpa
00:25 flaviu joined #marpa
00:32 ronsavage joined #marpa
00:49 ronsavage Re my editor problems. They've been resolved by installed another jpeg library. Whew!
01:15 Spads_ joined #marpa
01:51 koo6 joined #marpa
01:56 jeffreykegler joined #marpa
03:57 Spads_ joined #marpa
04:25 ronsavage joined #marpa
05:35 rns_ joined #marpa
05:42 rns_ jeffreykegler: re http://irclog.perlgeek.de/marpa/2014-11-19#i_9689889 — Actually, I think we should. I have liked the Perl6 approach since http://irclog.perlgeek.de/marpa/2014-11-19#i_9685669 :)
05:44 rns_ Still, Ron offered a nice back-and-forth and I chimed in.
05:47 rns_ ronsavage: re http://irclog.perlgeek.de/marpa/2014-11-19#i_9689902 — yes, but as Jeffrey says, brackets are much more context-sensitive.
07:16 ronsavage OK. Perl6 syntax it is. We just have to hope Perl6 does not fade away and leave us as the odd-man-out!
08:12 Spads joined #marpa
08:29 lwa joined #marpa
08:46 jeffreykegler joined #marpa
08:49 jeffreykegler re http://irclog.perlgeek.de/marpa/2014-11-20#i_9691469 -- If Perl 6 fizzles, and the LUIF preserves some of its insights, that would be a fine thing.
08:52 jeffreykegler As far as folks coming to the LUIF who are already familiar with Perl 6, I don't think there will be a lot of those.  But the syntax I'm stealing is for problems that really nobody else but Larry and Marpa have tackled, so nothing else will be more familiar.
08:53 jeffreykegler In the early days, Larry envisioned Perl 6 as a kind of language-building toolkit, and I think to a certain degree still does.
08:58 jeffreykegler But Perl 6 bases these hopes on a top-down parsing approach and how likely I think that is to work, I think you can see from my most recent blog post -- http://jeffreykegler.github.com/Ocean-of-Awareness-blog/individual/2014/11/ll.html
09:27 ronsavage Yes - I don't see hoards of programmers moving from Perl6 to Marpa, but at least their regexp knowhow in Perl6 will ease the transition. And moving the other way will be eased, too.
09:28 ronsavage I'm glad we - even if briefly - discussed alternatives, and you're right. Larry has thought about this for so long, we're actually lucky to pigglyback on that.
10:29 rns_ joined #marpa
10:29 rns_ re Perl6 — The last pieces are finally falling into place. After years of design and implementation, 2015 will be the year that Perl 6 officially launches for production use.
10:30 rns_ https://fosdem.org/2015/schedule/event/get_ready_to_party/ — to be spoken by Larry Wall.
10:40 daxim loup vaillaint's implementation notes say to use dynamic arrays as the appropriate data structure for earley/table-driven parsers, and indeed his lua example code uses lua tables (which grow dynamically), and nearley uses javascript arrays (dito)
10:42 daxim what data structure does R2 use in its C core?  in the distro tarball, where do I have to look to see the implementation?
10:52 daxim marpa really needs EBNF.   I'm tired of manually translating things like   ( foo ){3,5} bar* quux?  into BNF.  it's error-prone, and such expanded BNF rules become unreadable very quickly.    EBNF out of the box is the only thing nearley does better than marpa, but that makes it already ten times more convenient to work with.
11:00 rns_ joined #marpa
11:06 rns_ re http://irclog.perlgeek.de/marpa/2014-11-20#i_9692271 — {n,m} quantifiers for character classes (not really sure about sequences) is what I think can be useful.
11:07 daxim if you can see it's useful for character classes, why are you not sure about sequences?
11:09 daxim when would you ever prefer "foo foo foo | foo foo foo foo | foo foo foo foo foo" over "(foo){3,5}"?
11:10 rns_ For one thing, because standard EBNF doesn't have them — http://en.wikipedia.org/wiki/Extended_Backus%E2%80%93Naur_Form#Table_of_symbols
11:13 rns_ For another, if BNF is perceived as a set of rewriting rules, then I caught myself thinging that, e.g.
11:13 rns_ hr_indentation ~ ' ' | '  ' | '   '
11:13 daxim alright, apparently I'm talking about ABNF
11:15 rns_ ... as I had to write for CommonMark https://github.com/rns/MarpaX-Languages-CommonMark-AST is more readable in terms of providing clearer rewriting options.
11:16 daxim where did you get that grammar from? did you derive it yourself from the spec prose text?
11:20 rns_ Yes. The authors didn't see it worth the effort to write the grammar because they knew (or thought so) that there can be no parser for it. But Marpa (probably with procedural parsing via events) can do it.
11:22 daxim awesome
11:26 rns_ Glad you like it. :) Sorry, have to get back to $work.
11:26 rns_ AFK
11:43 jluis joined #marpa
14:16 koo6 joined #marpa
18:14 jluis joined #marpa
18:41 idiosyncrat joined #marpa
18:43 idiosyncrat daxim: Libmarpa is written in C and the Earley tables uses linked lists with some dynanmtic arrays, for example for indexing the list items by symbol -- it's complex
18:44 idiosyncrat I was going to refer you to the Libmarpa web page, but I discover it's hopelessly out of date.  So I'll fix it and update it and let the IRC channel know.
18:45 idiosyncrat Libmarpa now lives in its own Github repo: https://github.com/jeffreykegler/libmarpa
18:46 idiosyncrat Realistically, it's best to start with the pseudo-code in the theory paper, because the paper contains a lot of explanations which are not repeated in the code or its comments.
19:16 koo6 joined #marpa
19:30 jdurand_ joined #marpa
19:31 jdurand_ Re  http://irclog.perlgeek.de/marpa/2014-11-20#i_9692271 I personnally have a full EBNF implementation that understand perfectly the W3C's BNF - which is an EBNF thingy
19:39 jdurand_ the only difficuly is the exclusion rule A ::= B - C that is not builtin but has to faked in user space
20:15 idiosyncrat The Libmarpa web page is now rewritten, and at this point what it says should have some basis in reality.
20:15 idiosyncrat http://jeffreykegler.github.io/Marpa-web-site/libmarpa.html
20:16 idiosyncrat daxim: it answers your questions about finding and reading Libmarpa.
20:36 flaviu joined #marpa
20:42 idiosyncrat joined #marpa
20:59 Spads joined #marpa
20:59 Spads joined #marpa
21:35 Spads joined #marpa
21:44 ronsavage I've updated Marpa's homepage to include MarpaX:: Languages:: CommonMark:: AST because I was not previously aware of it. If you notice anything else missing, please let me know.
22:11 jeffreykegler joined #marpa
22:48 lwa ronsavage: It might be worth noting in that list which language interfaces are still works in progress or incomplete.
22:48 lwa For example, the CommonMark implementation currently seems to ignore nesting blocks (e.g. in lists or blockquotes), which is precisely the feature that makes block-level Markdown impossible to describe as a context-free language, and extremely difficult to parse via Marpa.
23:49 jeffreykegler joined #marpa

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