Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-08-10

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

All times shown according to UTC.

Time Nick Message
01:30 ronsavage joined #marpa
01:31 ronsavage Marpa::R2 V 2.089001. Test statistics:
01:31 ronsavage Fails: 0. Files: 14. Modules: 8. Passes: 8. Tests: 523.
01:31 ronsavage Duration: 1 minute and 25 seconds
01:47 ilbot3 joined #marpa
01:47 Topic for #marpa is now IRC logs: http://irclog.perlgeek.de/marpa/today - Pastebin: http://scsys.co.uk:8002/marpa - Released code: https://metacpan.org/pod/Marpa::R2 - Source code: https://github.com/jeffreykegler/Marpa--R2 - Mailing list: https://groups.google.com/forum/?hl=en&fromgroups#%21forum/marpa-parser - Web site: http://jeffreykegler.github.io/Marpa-web-site/
04:17 jeffreykegler joined #marpa
04:18 jeffreykegler ronsavage: re http://irclog.perlgeek.de/marpa/2014-08-10#i_9159851 -- Thanks!
15:00 jeffreykegler joined #marpa
21:45 jeffreykegler joined #marpa
21:47 jeffreykegler An irony re Lua + Marpa -- Lua's parser is recursive descent -- and very arguably it is the right choice.
21:48 jeffreykegler Lua is unusual in that it has a very small grammar and scalability does not matter -- it is a permanent design goal of Lua *not* to expand its syntax.
21:49 jeffreykegler Recursive descent's big problem is that for a grammar of any size, all that custom code is a rat's nest, one that gets worse as it scales.\
21:49 jeffreykegler But for Lua this problem is small, and is guaranteed to stay small.
21:50 jeffreykegler Ordinarily Marpa would still have the advantage, because usually "small and staying small" == "quick and dirty"
21:51 jeffreykegler So Marpa's faster development time would be the overwhelming advantage.
21:52 jeffreykegler But Lua, while small and staying small, is also production quality -- it's very far from quck and dirty.
21:53 jeffreykegler Even in this unusual case, the trade-offs are close, but a reasonable person could prefer the recursive descent approach for Lua.
21:54 jeffreykegler Recursive descent allows (in fact requires) custom hacking, and Lua's case is one of the rare ones where custom hacks do not get out of hand due to scale, and also have a real justification in terms of the design goals.
22:35 jeffreykegler Summarizing the above: recursive descent makes sense if it makes sense to spend a lot of time carefully hand-crafting a parser that is targets a small language, and that will not scale.
22:37 jeffreykegler Lua, as it happens is such a case.  It's small.  It's important to squeeze the last ounce of performance out of the parser, and worth time and trouble to do it.  And the language will stay very small.  Unscalable in this context is perfectly OK.
23:18 ronsavage joined #marpa

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