Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-10-11

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

All times shown according to UTC.

Time Nick Message
00:27 flaviu joined #marpa
00:53 jeffreykegler I forked rns's Lua parser and added my first draft of the LUIF BNF statement to it: https://github.com/jeffreykegler/MarpaX-Languages-Lua-AST/commit/f13edc89ac3b6c290e0a4f28b8274927ec3c967f
00:58 jeffreykegler When I ran the test suite, some fail, but I think the same ones both before and after my changes.
00:59 jeffreykegler In this form, it's perhaps less than 100% clear where I'm going, but once we can turn some examples into parse trees, that will help.
01:00 jeffreykegler Also, I have not created a pull request.  If we want to do a Lua formatter, we may want to fork LUIF away from the 'pure Lua' parser
03:19 CQ_ joined #marpa
03:48 nearley joined #marpa
05:13 qu1j0t3 joined #marpa
06:41 lwa joined #marpa
08:54 ronsavage joined #marpa
10:28 rns joined #marpa
11:19 flaviu joined #marpa
12:17 rns joined #marpa
12:18 rns jeffreykegler: Does this commit http://goo.gl/TojdoR fixes a bug in sl_external1.t from Marpa::R2 test suite (the code was borrowed from there)?
12:18 rns For me it does, but perhaps I'm missing something?
12:22 rns jeffreykegler: re http://irclog.perlgeek.de/marpa/2014-10-11#i_9489232 — I added and extension mechanism to Lua parser, e.g. http://goo.gl/ORsW8W
12:24 rns The point is, I've moved the parser to external lexing recently so extension by editing the grammar is not so easy anymore. :)
12:27 rns This was the first cut, now — http://goo.gl/wzvH3M — I use synopsis code from Marpa::R2.
12:31 rns Some possible use of extensible parser would be smart preprocessing, e.g. testing syntactic sugar or forbidding some syntax to ensure adherence to some version.
12:32 rns Perhaps the lua community would find it useful.
12:33 rns Also on extension front, but more about extensible lexing: can you look at this gist http://goo.gl/C8gIe5 when tike permits?
12:34 rns s/tike/time/.
12:35 rns It would allow (or so I think) adding not only string, but also regex lexemes to external lexer arbitrarily, because the lexer can infer the token from the matched value.
12:35 jeffreykegler joined #marpa
12:36 rns Now new literals must be passed explicitly to extend() as shown in the extend.t.
12:37 shadowpaste "jeffreykegler" at 162.232.214.245 pasted "Problem with new Lua::Ast compiling" (6 lines) at http://scsys.co.uk:8002/429231
12:38 jeffreykegler Actually, the latest Lua::Ast bails for me on the first test.  I just pasted the error I get.
12:38 rns is it from freshly cloned repo?
12:39 rns Oh sorry, missed "the latest" I'm looking into it.
12:39 jeffreykegler "Latest" but *not* freshly cloned.
12:39 jeffreykegler I'll try that.
12:41 rns Well, I'm doing it now; I run prove from the editor so there can be issues. Lemme try to reproduce it.
12:43 jeffreykegler OK, freshly cloned shows same problem.  I'm going back to sleep now.  AFK!
13:24 flaviu joined #marpa
13:24 rns jeffreykegler: Should be (I hope) fixed in this commit http://goo.gl/rhBdQR — I had shebang in the module file. Sorry if I woke you up — didn't want to.
13:42 rns Freshly cloned repo tests better now — thanks for pointing me to this.
13:42 rns left #marpa
13:54 flaviu joined #marpa
14:39 rns_ joined #marpa
14:47 qu1j0t3 left #marpa
14:59 rns_ LUIF progress so far: Marpa:R2 synopsis code parses without actions and fails with them (1 test).
14:59 shadowpaste "rns_" at 77.120.243.111 pasted "LUIF rules with actions test" (85 lines) at http://scsys.co.uk:8002/429238
15:00 rns_ left #marpa
15:02 rns joined #marpa
15:03 rns I need to go AFK till next morning (for 10-12 hours), but will backlog.
15:03 rns left #marpa
21:38 jeffreykegler joined #marpa
21:40 jeffreykegler rns: still having a few troubles running Lua-AST.  You should add version numbers for perl and the modules needed.
21:40 jeffreykegler Because I support 5.10, I try to always run as 5.10, so I really need the heads-up if a later Perl is needed.
21:53 jeffreykegler Having the version numbers, it fails fast and with a clearer error message
23:11 jeffreykegler rns: re http://irclog.perlgeek.de/marpa/2014-10-11#i_9491114 -- the problem is that <action literal> is not being lexed as anything special, just as <Name>
23:14 sivoais joined #marpa
23:30 jeffreykegler rns: Filed a pull request with some other changes, including a fix to my LUIF grammar
23:34 jeffreykegler By the way, I don't know how it will work out, but what rns and I are dolng here is a first -- language design starting with the parser.
23:35 jeffreykegler With recursive descent, this is clearly not possible.
23:36 jeffreykegler With yacc, in theory it'd be possible, but in practice getting the parser working is so hard, you can't start there, but have to have the language pretty completely designed first.
23:36 jeffreykegler Here it's "Let's write the parser and see what the language looks like".
23:37 jeffreykegler Once the parser is ready, I intend it, and the language it both specifies and implements, as a basis for discussion.
23:38 jeffreykegler (Again, something you can't really do with yacc.  With yacc, once you've gotten the parser working you breathe a sigh of relief and hope you'll never have to do anything with it ever again.)
23:45 jeffreykegler It's sort of "RFC by implementation"

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