Perl 6 - the future is here, just unevenly distributed

IRC log for #rakudosketch, 2010-03-31

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

All times shown according to UTC.

Time Nick Message
01:41 sorear joined #rakudosketch
18:53 ash_ joined #rakudosketch
18:57 mberends joined #rakudosketch
19:06 TimToady not that anyone else is here...
19:06 jnthn 'tis a tad quiet.
19:06 moritz_ nobody
19:07 * moritz_ completely forgot
19:07 jnthn Great, that's three of us not here. Any more?
19:07 jnthn ;-)
19:07 jnthn moritz_: Same!
19:07 ash_ i am more just listening, since i don't really commit to rakudo
19:07 moritz_ anyway, last week I promised to look at 1) tests for item assignment and 2) match object perl6ification
19:08 moritz_ for 1) I fudged assign.t for rakudo, runs now
19:08 moritz_ plenty of item assignment tests
19:08 moritz_ 2) I started a branch on nqp-rx that splits up match object creation
19:08 moritz_ it does all the logic in the Regex::Cursor object
19:09 moritz_ and then calls a Regex::Match.new with all the necessary information, including submatches
19:09 jnthn moritz_++
19:09 moritz_ so that providing a different type object makes it easy to create custom Match types
19:10 moritz_ however I want review from pm before I merge
19:10 jnthn OK, cool.
19:10 jnthn Is the next stage getting it to use some Rakudo match object instead?
19:10 jnthn Sounds like a clean solution, anyway.
19:10 moritz_ aye, I plan to write class Match in pure Perl 6
19:11 moritz_ and nqp-rx will just call its .new method
19:11 jnthn Excellent.
19:12 jnthn I did some little patches here and there and Book work.
19:12 jnthn Need to jump back on the OO things, and the multi things, and the lexical things...
19:13 jnthn Lexical multis are muchly needed.
19:13 jnthn Thinking about how we can achieve a "real" setting.
19:13 jnthn (as in, lexical outer)
19:14 jnthn A little non-trivial in various ways.
19:14 moritz_ care to explain what the difficulties are?
19:15 moritz_ I'd expect you "just" need a function to iterate over lexpads, and import that stuff into another scope
19:15 jnthn Well, we can do it that way too.
19:15 jnthn I was wondering also if we could have a "you are here" marker like STD uses.
19:15 jnthn And set_outer the main body of the program.
19:16 jnthn But the trouble with that is then getting the parser to know about all of the stuff in said outer scope in a neat way.
19:16 jnthn Oh
19:16 jnthn We may not need to actually.
19:16 jnthn Since we emit a "lexicals first, then namespace" op for sub lookup.
19:17 jnthn So maybe it ain't going to be so bad after all.
19:17 TimToady you have to make sure that OUTER:: works correctly
19:17 jnthn Apart from actually the parser probably will need to know in some senses, for static analysis.
19:17 jnthn But we don't do that now, so...no biggie.
19:17 TimToady UNIT::OUTER == SETTING, and so forth
19:18 TimToady well, that's part of what it means to be the outer lexical scope
19:18 jnthn nod
19:19 jnthn Maybe I should fix lexical multis and JFTI.
19:20 moritz_ sounds like a plan
19:23 jnthn Anyways, things seem to be moving along overall.
19:24 mberends when refactoring proto, the nested project folders made by Rakudo's --gen-parrot looked potentially awkward for future re-structuring, so I am going for default directories of ~/.perl6/parrot, ~/.perl6/parrot_install and ~/.perl6/rakudo. Any comments on that?
19:25 jnthn mberends: Modifying the default behavior of --gen-parrot?
19:25 mberends no. just proto
19:25 jnthn Ah, OK
19:25 jnthn Will things still work out for a Rakudo not installed with proto?
19:26 mberends one underlying thought was to remove parrot/ and rakudo/ and retain only parrot_install for runtime.
19:26 jnthn Hmm
19:26 jnthn +1 if that works.
19:26 mberends jnthn: yes, if you byo Rakudo proto will use that instead
19:27 jnthn Is there or will there a way to get proto to do an install to /usr/bin/local too (e.g. a global rather than user one)?
19:27 mberends you need root access on Unix for that
19:28 jnthn Oh, sure. :-)
19:28 mberends that's why ~/.perl6 seemed a safer choice
19:28 jnthn I can just see people wanting it and was wondering what the scope of proto installing Rakudo/Parrot was.
19:28 jnthn I kinda worry a little that there's perl Configure.pl --gen-parrot and then the proto way, and how we'll guide people which to use.
19:28 mberends proto could include an optional sudo step to install into /usr/local/bin
19:28 ash_ mberends: you could detect if the current user is an admin, do a test to see if you have permissions to /usr/local/bin then if that fails, fallback to ~/.perl6/lib
19:29 jnthn Especially given proto will ship with Rakudo *.
19:29 mberends ash_: no problem, that could be added
19:30 ash_ there are other package management systems that follow that scheme, if you don't have permissions to install it to the globally accessible place, then fallback to ~
19:30 ash_ its kinda nice
19:30 mberends yeah, I like it. CPAN also does that kind of thing
19:31 mberends ok, there being no dissent, the proposal is carried
19:31 spinclad [bikeshed] i'd rather ask specifically to install in /usr[/local] and fail if no perms
19:32 spinclad [too late]
19:32 jnthn But purple is better!
19:32 jnthn :-)
19:32 * moritz_ likes locality
19:32 spinclad anyway, no blocker for R*
19:33 mberends spinclad: no problem either, I'm going to build the guts, and we can orchestrate the directories by tweaking a few high level parameters
19:33 mberends there is a proto.conf for those kind of options, and favorite colors ;)
19:34 colomon joined #rakudosketch
19:36 mberends this afternoon Rakudo 2010.02 failed to compile one .c file with Strawberry's g++ 3.4.5, I'm not sure of the details, it may have been bind.c :(
19:37 jnthn g++?
19:37 ash_ g++ 3.4.5 is from 2005....
19:37 jnthn Feeding C code to a C++ compiler may not always end well. ;-)
19:37 ash_ gnu c++ compiler
19:38 mberends ahem, yes. That's in Strawberry 5.10.1, current (January 2010) release.
19:40 mberends ActivePerl is beginning XS support with a gcc package as well, haven't tried it yet
19:43 mberends anyway, these details are more appropriate for #perl6. After I get proto minimally operational, I plan to switch my efforts back to Blizkost/FakeDBI
19:52 * spinclad senses quiet
19:56 mberends re-testing, bind.c was the last file to compile ok. Configure had nominated g++ for $(LD), and it failed to link src\pmc\perl6_group.dll with "undefined reference" errors
19:59 ash_ mberends: can you put the whole error and the command it executed in a gist or something?
20:00 TimToady phone &
20:00 mberends ok, will post in #perl6
20:32 PerlJam joined #rakudosketch
20:33 colomon Was there a meeting while I was out?
20:34 jnthn Sorta
20:34 mberends not much, just see http://irclog.perlgeek.de/rakudosketch/today
20:34 jnthn There's ir clogs
20:34 colomon danke
20:36 colomon yeah, very short.  :)

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