Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-01-21

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

All times shown according to UTC.

Time Nick Message
04:11 jeffreykegler I've just posted to the Google group re licensing.  I'm asking about --
04:12 jeffreykegler changing the Libmarpa license to the MIT license; and
04:13 jeffreykegler adding language that assigns copyright of all contributed code to me -- I discuss the implications of this, pro and con, in the post.
04:55 ronsavage joined #marpa
05:55 ronsavage jdurand: Not released, but available: savage.net.au/Perl-modules/Text-Delimited-Marpa-1.00.tgz. It handles nested delimited strings as you mentioned. See the Description and the 1st two QA in the FAQ for differences with Text :: Balanced :: Marpa. A new version (1.07) of the latter is being uploaded to CPAN now. There are only minor changes.
06:48 basiliscos joined #marpa
07:26 jluis joined #marpa
08:12 koo5 joined #marpa
08:42 ronsavage Everyone: Please proofread the docs for the new module: http://savage.net.au/Perl-modules/html/Text/Delimited/Marpa.html
09:16 rns Typos caught are in this PR -- https://github.com/ronsavage/Text-Delimited-Marpa/pull/1.
09:16 rns Perhaps that's only me, but I cannot fathom "This message can never be just a warning message." Is it always fatal? Or it means "you should call "error_number()" to find out what happened?"
09:16 rns Links in places like "See t/escapes.t", if possible, wouldn't hurt.
09:30 rns ronsavage: And thanks for acknowledging my actually small contribution. Never thought it could evolve so usefully, a good thing to see.
12:29 basiliscos joined #marpa
12:31 basiliscos joined #marpa
12:33 jeffreykegler joined #marpa
12:33 jeffreykegler ronsavage: Wonderful stuff, thanks!
12:34 basiliscos joined #marpa
13:16 koo5 joined #marpa
13:50 btyler joined #marpa
15:31 lwa joined #marpa
16:03 basiliscos joined #marpa
16:06 basiliscos joined #marpa
16:09 basiliscos joined #marpa
16:12 basiliscos joined #marpa
16:15 basiliscos1 joined #marpa
16:17 basiliscos joined #marpa
16:20 basiliscos joined #marpa
16:25 basiliscos joined #marpa
16:28 basiliscos joined #marpa
16:31 basiliscos joined #marpa
16:34 basiliscos1 joined #marpa
16:37 basiliscos joined #marpa
16:39 basiliscos joined #marpa
16:42 basiliscos joined #marpa
16:46 basiliscos joined #marpa
16:48 basiliscos joined #marpa
16:51 basiliscos1 joined #marpa
18:04 basiliscos joined #marpa
19:22 jdurand_ joined #marpa
19:24 jdurand_ basiliscos: have you successfully tested latest cscan - c.f. https://github.com/jddurand/MarpaX-Languages-C-AST/issues/11 ? Thx.
19:25 basiliscos jdurand_: yes. Thats one works perfect!
19:34 jdurand_ ok thnx. I'll fix the test suite that recently failed on solaris only and will push an indexed release
19:48 jdurand_ ronsavage: Re http://irclog.perlgeek.de/marpa/2015-01-21#i_9977263 - Thx! Is it supposed to parse something like e.g. /* This is a /* test */ where open delimiter is '/*' and end delimiter is '*/' ?
19:55 jdurand_ ... or perhaps this: /*/*/ - I ask because I am /not/ sure this really belongs to "delimited match" notion - just puzzling cases -;
20:02 jdurand_ basiliscos: would you mind to resume to us in which context/project you plan to use cscan ?
20:03 basiliscos jdurand_: sure
20:04 basiliscos wait a little bit
20:10 basiliscos there is a project - clusterGL http://clustergl.org/ - it is used for "distributed openGL rendering"
20:11 basiliscos the idea is rather simple - instead of directly dispatch opengl commands, they capture them, serialize, send to several nodes (displays) on network, and execute opengl commands on the final nodes
20:12 basiliscos to get display wall they to a little trick: they "mangle" glViewPort command to for each node to display appropriate view of the whole scene on the particular node
20:13 basiliscos - that's good. The bad part, is that: they already generated capture/serialization/deserialization/execution for opengl commnds (over 1.5k)
20:14 basiliscos - and then they did manual hacking of the generated code.
20:14 basiliscos - so it works OK for quake1 and may be other stuff, but not for my simple opengl commands
20:15 basiliscos So, I decided to rewrite it from scratch more clever way: https://github.com/basiliscos/cpp-clustergl-ng
20:16 basiliscos - leave generated code as it is is, i.e. untouched,
20:16 basiliscos but if there is manually re-written openg-command (serializer/capturer/executer etc.), - it should be taken instead of generation
20:17 basiliscos for example for glViewPort
20:18 basiliscos so, (1) I use cscan to get list of opengl functions with params, with further generation of all required wrappers
20:18 basiliscos and (2) I plan to analyze the source, and if there is present human-written wrapper for particular command, then no generation should occur
20:19 basiliscos that is the story
20:19 basiliscos but I'm still on the early stages :)
20:24 rns left #marpa
20:24 jdurand_ That's very nice and interesting. Thanks a lot for the explanation. When your project will be ready for testing, I hope the reader will pay attention not only to your nice work, but also to the cscan tool - which I believe is truely powerful
20:25 rns joined #marpa
20:26 basiliscos indeed. I hope all it should be build from scratch, i.e. no generated files should be pre-shipped.
20:28 basiliscos One drawback - that it takes a some time... for building AST XML . About 10 minutes for gl.h on my budget notebook. (And there will also be glext.h).
20:28 jdurand_ You're talking to somebody that is a fan of code generation -; !
20:28 basiliscos Not a huge problem, unless you are gentoo user :)
20:29 jdurand_ Yes... AST takes time sometimes. I may rewrite that OOTD, but is lacking time - though the current version works OK. Just a bit slow.
20:29 basiliscos jdurand_: I already "spread a word" on StackOverflow about cscan: http://stackoverflow.com/questions/239722/ast-from-c-code/27824040#27824040  )
20:31 jdurand_ Oh!!! I upvoted lol
20:35 jdurand_ Ok... MarpaX-Languages-C-AST-0.44-TRIAL uploaded - with tricky hacks to get around different behaviour of cpp v.s. space on the different OSes... For those interested I am thinking to something like e.g. http://www.cpantesters.org/cpan/report/6218cf7c-a0da-11e4-b221-9e126cbd7f71
20:46 jdurand_ Ron: impossible for me to sign in to blogs.perl.org, even with password change..; Just wanted to say "nice work" about X500::DN::Marpa
21:03 ilbot3 joined #marpa
21:03 Topic for #marpa is now Start here: http://savage.net.au/Marpa.html - Pastebin: http://scsys.co.uk:8002/marpa - Jeffrey's Marpa site: http://jeffreykegler.github.io/Marpa-web-site/ - IRC log: http://irclog.perlgeek.de/marpa/today
21:06 ronsavage jdurand: Re http://irclog.perlgeek.de/marpa/2015-01-21#i_9981320. Thanx!
21:10 jdurand_ joined #marpa
21:11 jdurand_ Opened question: how can I address the best lats comment at https://github.com/jddurand/MarpaX-Languages-SQL2003-AST/issues/1 ? Thx.
21:11 ronsavage jdurand: Re http://irclog.perlgeek.de/marpa/2015-01-21#i_9980931. scripts/synopsis.pl uses /* and */. So, yes, it's meant for that. Or did you mean when the quotes are unbalanced, as in /* One /* Two */? ATM the unbalanced case is a fatal error. That can be changed. This type of issue is why I did not release the module. I want feedback on how you think it should operate.
21:12 jdurand_ ronsavage: Hmm... I see the point... don't know yet... I guess your module reacts correctly, though I'm awaiting others feedback
21:13 jdurand_ Still, many thanks for your excellent recent modules
21:13 jdurand_ AFK!
21:14 ronsavage jdurand: Re http://irclog.perlgeek.de/marpa/2015-01-21#i_9981545. Can't you replace <SQL_Start_Many> with <SQL_prefix>, which is defined with 2 rules:
21:14 ronsavage <SQL_prefix> ::= <Token> <SQL_Start_Many>
21:14 ronsavage <SQL_prefix> ::= <SQL_Start_Many>
21:50 jeffreykegler joined #marpa
22:55 ronsavage joined #marpa

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