Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-07-25

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

All times shown according to UTC.

Time Nick Message
00:34 jeffreykegler joined #marpa
00:41 jeffreykegler jdurand: re http://irclog.perlgeek.de/marpa/2014-07-24#i_9076639
00:42 jeffreykegler It's not going to be Kollos vs. Perl -- R3 is going to be Kollos-based.  From Perl's point of view, Lua is all that semantic code I would have had to write myself -- except that it's more fully featured and has a ready-made support community.
00:44 jeffreykegler Because of the way Lua is designed, the additional overhead of going this way is zero or negative -- negative because Ierusalimsky is very focused on size and efficiency, is very smart about it, and is able to do it full time, whereas I'd have to do it as a side show.
00:45 jeffreykegler R2 is probably going to move from "under active development" to "stable".  Bug fixes, and enhancements which has a good risk-reward ratio for existing users will continue to happen -- one of these is support for the Debian packaging.
00:47 jeffreykegler Over the past year, I've managed to keep R2 both stable, and under very active development, with full backward compatibility ---
00:48 jeffreykegler that has gone well past the point of diminishing returns.
00:50 jeffreykegler A very real hope I have for Kollos, is that this will allow others to contribute in the middle layers, the layers between the applications and the Libmarpa parse engine, where THIF, SLIF, NAIF, etc. live.
00:50 jeffreykegler I hope that answers the questions. :-)
03:03 ronsavage joined #marpa
04:26 jeffreykegler joined #marpa
04:27 jeffreykegler To give a bit of perspective on Kollos (Marpa + Lua)
04:28 jeffreykegler Originally, Marpa's semantics are Perl -- via callbacks.  But that's slow, so I started writing shortcuts, like ::first for "select the first child" and ::array for "put all the child nodes into an array"
04:29 jeffreykegler With the shortcuts you get speed.
04:30 jeffreykegler (For the ultimate in speed, you call Libmarpa directly from C, but then you give up all the conveniences Perl offers.)
04:31 jeffreykegler The shortcuts give you close to the best of all worlds -- but they do very specific things.
04:32 jeffreykegler The obvious next step is write more shortcuts, allow conditionals; throw in some arithmetic.  How about variables while you're at it?
04:35 jeffreykegler This, of course, amounts to growing a new language "organically" -- that is, opportunistically, and without planning.  This is the way the vim macro language, JCL and the "Bourne shell language" came about.  It's a recipe for disaster.
04:37 jeffreykegler So this is where Lua is coming in.  It replaces all those special action shortcuts with a nice, well-planned, small, tested and supported mini-language which is very close to the metal.
04:39 jeffreykegler Perl is still there, and you'll get your choice of the big expensive language; or the efficient, if sometimes infurating, small language -- or you can mix them.
06:42 ronsavage joined #marpa
09:01 aredridel joined #marpa
10:15 ronsavage joined #marpa
10:16 ronsavage This all sounds good to me.
15:30 jeffreykegler joined #marpa
17:54 jeffreykegler1 joined #marpa
22:27 jeffreykegler joined #marpa
23:38 ronsavage Hahaha. Someone else who is far from impressed with vim's macro language. I understand people can force it to provide a myriad of shortcuts, but the syntax!

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