Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-10-25

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

All times shown according to UTC.

Time Nick Message
00:23 ronsavage jdurand: MetaCPAN is offline for me, and MarpaX :: Languages :: ECMAScript :: AST won't install: https://gist.github.com/ron​savage/8ba445d1fcad12d7c53b. Any ideas? TIA.
00:46 flaviu ran marpa on a large file and measured memory usage: https://gist.github.com/ec077257d7b362f12dc4
00:46 flaviu I'm not asking for assistance, just posting that. I just ran it to see what it could handle.
03:43 jdurand joined #marpa
03:46 jdurand Re http://irclog.perlgeek.de/​marpa/2014-10-25#i_9563571 - Starting with 5.18.1 something is wrong, no modification in the code - c.f. http://matrix.cpantesters.org/?dist=M​arpaX-Languages-ECMAScript-AST+0.018 - I have no idea for the moment
04:04 jdurand joined #marpa
04:04 jdurand Ah, "good" news - reproduced the install failure iwth perl-5.2.0.0 on debian - looking at it
04:32 jdurand Looks like a Marpa::R2 regression -never seen that error before: "Error in SLIF parse: Unrecognized problem code: unpermitted mix of external and internal scanning" - submitted as issue 208, c.f. https://github.com/jeffreyk​egler/Marpa--R2/issues/208
05:42 ronsavage joined #marpa
05:42 ronsavage jdurand: Thanx.
05:44 ronsavage However, I get this Marpa exception error (on 2 lines): Exception as string is "Bad quantifier {6,4} in regular expression"
05:44 ronsavage Marpa::R2 exception at /home/ron/.cpanm/work/1414196041.15304/Mar​paX-Languages-ECMAScript-AST-0.018/blib/li​b/MarpaX/Languages/ECMAScript/AST/Impl.pm line 76.
05:44 ronsavage Did you see something different? I'm using Perl 5.18.2.
05:49 jdurand joined #marpa
05:50 jdurand ronsavage: ""Bad quantifier {6,4} in regular expression" is a predicted failure, string in stderr is normal, and the test is ok
06:06 ronsavage jdurand: OK.
06:57 rns joined #marpa
07:00 rns ronsavage: re http://irclog.perlgeek.de/​marpa/2014-10-24#i_9563149 — absolutely welcome.
07:00 rns A number of my gists — https://gist.github.com/rns — are marked as stackoverflow, so perhaps you can find smth. useful among them.
07:03 rns jeffreykegler: re Lua::AST — https://github.com/rns/MarpaX-Langu​ages-Lua-AST/blob/master/t/extend.t
07:04 rns contains the first cut at translation of stat :: BNF into pure lua.
07:05 rns I took the liberty of adding a rule to group BNFs under a single name, like this:
07:05 rns BNF ::= <BNF literal> Name <BNF rules>
07:05 rns <BNF rules> ::= <BNF rule>+
07:06 rns The result is a lua table, e.g. https://github.com/rns/MarpaX-Language​s-Lua-AST/blob/master/t/extend.t#L408
07:07 rns The rule adverbs (e.g. quantifier/separator for sequence rules, priority for prioritized rules) are in fields for now, together with actions.
07:08 rns The lua table layout is
07:08 rns lhs = { 'rhs1', 'rhs2', ... { adverb => value } }
07:11 rns I'd appreciate some early feedback. extend.t can be used as a base for a preprocessor (luamarpa?) to trasnpile luaBNF (lua with BNF rules) to pure lua.
07:12 rns Also, re http://irclog.perlgeek.de/​marpa/2014-10-24#i_9562431 (node.js folks interest in Marpa) —
07:13 rns perhaps they can be interested in LUIF, because they can use the same approach — there is a working ECMAScript parser
07:15 rns thanks to Jean-Damien — https://metacpan.org/pod/Marpa​X::Languages::ECMAScript::AST — which can be extended with BNF rules like we do with lua and perhaps even share some code interfacing lua to libmarpa.
07:16 rns And, last but not least :), re http://irclog.perlgeek.de/​marpa/2014-10-24#i_9562337
07:18 rns Marpa-R2-2.097_003 installed and works fine for Lua::AST and friends with perl5.20.1/vs 2010 and perl 5.14.2/gcc4.8.3 under winxp.
07:19 ronsavage rns: Thanx.
08:11 lwa joined #marpa
10:47 flaviu joined #marpa
11:11 flaviu joined #marpa
16:37 jdurand joined #marpa
17:35 jeffreykegler joined #marpa
17:38 jeffreykegler flaviu: Thanks for running those.  Right now, as this day's backlog shows, I've got a lot of other stuff, but I do have ways of addressing memory usage in mind, and at the point I get to that, the numbers will be very, very useful.
17:43 jeffreykegler Bottom line, when it comes to memory, Marpa is like DOM, and not like SAX, but there are ways in which a grammar could be parsed and processed a piece at a time.
17:43 jeffreykegler But I won't be getting to that any time soon.
17:53 flaviu Of course, I only ran that out of curiosity. I don't see it as becoming a problem for me in the future.
17:55 jeffreykegler ronsavage: re MarpaX::Demo::SampleScripts -- Thanks a lot for doing this.
17:57 jeffreykegler jdurand: re http://irclog.perlgeek.de/​marpa/2014-10-25#i_9564039 -- yes, "unpermitted mix of external and internal scanning" is a new error message.  I have guesses, but I will look at this before commenting further.
18:11 jeffreykegler rns: re rule names.  One of the ways I hope to entice people to work on Kollos, is that they'll be part of shaping it.  So adding a new hack is not taking a liberty.
18:11 jeffreykegler I do have thoughts, questions and even some reservations, however: https://github.com/rns/Marpa​X-Languages-Lua-AST/issues/2
19:03 jeffreykegler On Hacker News: https://news.ycombinator.com/item?id=8506682
19:04 jeffreykegler The commenter states that a problem with Marpa's adoption may have been its choice of implementation languages: Perl and CWeb
20:51 ronsavage joined #marpa
21:26 ronsavage Re http://irclog.perlgeek.de/​marpa/2014-10-25#i_9565896. Remember the saying "No matter what they're saying, it's a people problem"? It's people who have hear about Marpa, and then have to choose to adopt it. Blaming the innocent bystander (software) is a cop-out.
21:51 jdurand ronsavage: thanks to Jeffrey patience, I got convinced this is a bug in MarpaX::Languages::ECMAScript::AST, c.f. https://github.com/jddurand/MarpaX​-Languages-ECMAScript-AST/issues/5
21:53 jeffreykegler jdurand: Many thanks for your tireless efforts with Marpa
22:10 jdurand jeffreykegler: well, shame on me when reading the spec, i was not writen explicitely that postfix is not postfix anymore if there is an automatic semicolon insert before it, probably because it was too much obvious -;
22:10 jdurand "it was not"
22:11 jdurand Anyway, fixed and being uploaded
22:16 jdurand ./.. with all lexeme_read() return value checked as you pointed out -;
22:25 jeffreykegler AFK
23:47 ronsavage I given up trying to install MarpaX::Languages::ECMAScript::AST. Dist::Zilla installed 52 packages, and adding the authordeps installed another 140 before it died.
23:49 ronsavage So I used cpanm --force with Dist::Zilla::PluginBundle::TestingMania, and commented on github, and then used --force on Dist::Zilla::Plugin::ContributorsFromGit. Now try again....
23:49 ronsavage "I given up" => "I've given up", but perhaps not quite.......
23:51 ronsavage dzil install hangs with '(reading log message from standard input)' This is the first time I've used Dist::Zilla, so the question is: Is that read from STDIN normal? How do I proceed?
23:52 ronsavage I typed 'Test' <Enter>. Nothing happened so I tried ^C and it exited.
23:55 ronsavage Ahh. ^D got a bit further. I now get 'fatal: Not a git repository (or any of the parent directories): .git', so I guess the *.zip I d/l-ed from github needs to be a fork with a real .git/ dir.
23:56 ronsavage PS I did comment on github about the failure of Dist::Zilla::Plugin::ContributorsFromGit, which others have had. It seem the author /may/ have Gone Missing.

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