% particle has left particle!~particle@c-24-19-3-148.hsd1.wa.comcast.net
% particle has joined #parrotsketch
% particl1 has joined #parrotsketch
% particle has left particle!~particle@c-24-19-3-148.hsd1.mn.comcast.net
% particl1 is now known as particle
% jhorwitz has joined #parrotsketch
% jhorwitz has left #parrotsketch
% rdice has joined #parrotsketch
% pmichaud has joined #parrotsketch
% kjs has joined #parrotsketch
% rdice has left rdice!~richard_d@CPE0014bfafbbd5-CM0011e6ecf48a.cpe.net.cable.rogers.com
% Coke has joined #parrotsketch
% allison has joined #parrotsketch
% Coke has left Coke!~coke@cpe-72-228-52-192.nycap.res.rr.com
% Coke has joined #parrotsketch
% barney has joined #parrotsketch
<Coke> Hello, everybody.
<pmichaud> Hello!
<barney> Hi
<kjs> hello
<Coke> I have two things to report, then we'll do our usual alphabetical dance of reporting.
% chromatic has joined #parrotsketch
<Coke> I will be cutting the release this month; we had a backup for allison but real life intervened. As I have no real life, I'll take this one. I have a call out for volunteers for release manager for '08; hasn't hit the list yet due to my new email setup, but I'm surred it will soon; in the meantime, anyone interested in being a RM for 08 let me know; commit bits and attention to detail required.
<Coke> s/surred/assurred/
<pmichaud> Coke -- I saw the call for volunteers
<Coke> Ah, exceollent, apparently I have more email issues than I thought. =-)
<Coke> We're up to 4 at the moment; like to get at least six so we're covered through june.
<Coke> That's it for me. on to our asciibetical reporting.
<Coke> allison, you're up.
<allison> - I've been working with Andy Armstrong and others clearing up a few platform issues with the new concurrency scheduler. We're getting closer, but debugging by email is slow-ish.
<allison> (As an aside, it's interesting, I was expecting Windows to be the tricky one, but turns out it's Mac OS 10.5.)
<allison> - I have some code for event handling with the new scheduler that I haven't checked in yet because of those issues.
<allison> - I've also been exchanging email with Marton Papp as he works on compiling Parrot on LLVM (llvm-gcc, LLVM's C interface).
<allison> - And I'm getting started on the multimethod dispatch PDD.
<allison> EOR
<barney> Worked on Eclectus, Scheme on Parrot in Scheme.
<barney> Converted examples/past/01-sub.pir to the PCT-PAST.
<barney> Coke: you can add me to the RM list
<barney> .eor
<Coke> I already had it that way, barney.
<chromatic> I found and fixed two nasty bugs.
<chromatic> I'm happy to re-up as an RM.
<chromatic> .end
<Coke> kjs?
<kjs> Worked on a compiler for Neko (nekovm.org), the language of the Neko VM, which tries to accomplish
<kjs> the same thing as Parrot, only it's much smaller, and doesn't have so much infrastructure (AFAICT).
<kjs> I didn't check it in; I don't want to add Yet Another Incomplete Language. If I get it working
<kjs> correctly, I'll add it. 
<kjs> .end
<Coke> particle?
<particle> ~ i stalled on some perl 6 stuff, and decided to take on punie to free my mind
<particle> ~ punie now converted to nqp, and passing more tests than before
<particle> ~ i'd love to see how far i can get before release day
<particle> ~ my mind is not yet free, i'm too excited about advancing punie forward
<particle> ~ hope to get back to perl 6... need to get it working before release
<Coke> particle: you've been very busy with punie, nice to see that converted over. Anything else?
<particle> did you see '.end' in my message?
<particle> should be a seven line report, just posted
<Coke> nope. last was "working before release"
<chromatic> Agreed.
<particle> hrmm, okay
<particle> ~ have i mentioned how powerful and flexible pct is now? pmichaud++
<particle> .end
<pmichaud> * More PCT improvements, some API cleanups
<pmichaud> * Been helping others with language implementations using PCT
<pmichaud> * Converted pynie to use PCT during my plane trips last week
<pmichaud> ** now 473 lines PIR, 809 lines Perl 6  (and nothing else).
<pmichaud> * Posted another journal article today outlining some recent Parrot achievements
<pmichaud> * article also points out a couple of RT tickets where contributions would be we
<pmichaud> lcome
<pmichaud> * writing more requested articles about pct and nqp
<pmichaud> * converted more of perl6 to get more tests passing
<pmichaud> * also implementing perl6 base object types that is much closer to the Perl 6 and Parrot specs than the numerous workarounds we had before
<pmichaud> * High priority:  RT#48467 is going to be a real blocker for a while
<pmichaud> .end
<Coke> spinclad, tene, tewk?
<spinclad> .eor
<Coke> integral, leo, perljam?
<Coke> k. pmichaud had a question...
<pmichaud> kjs: can we mark RT#48152 as resolved?  (missing .get_results documentation)
<kjs> I think I added it to PDD19
<kjs> ehm, so that would be yes :-)
<pmichaud> I saw that you did, so I'm thinking... yes.  Thanks.
<kjs> I'll close the ticket
<pmichaud> kjs++
<pmichaud> I have a design-ish question after we get through any other questions.
<allison> I have an other-ish question
<Coke> allison: go ahead.
<allison> pmichaud: wasn't RT#48467 (your blocker) a work around for the fact that objects couldn't be morphed?
<pmichaud> it was more of a workaround to get perl6's tests to pass
<pmichaud> however perl6's tests didn't test the specific case I ran across last night
<pmichaud> and that's likely to be a very common case
<pmichaud> (assignment, that is)
<pmichaud> again, "morphing of objects" isn't really the target -- I just want to have an assignment-like operation that mimics what happens in a hll
<pmichaud> (morphing could be considered a side effect of that assignment, yes.)
<allison> so is every assignment in Perl 6 a morph in Parrot?
<pmichaud> potentially
<pmichaud> my $x = 'hello';   $x = $x + 1;    # morphs $x
<allison> yeah
<pmichaud> my $x = [ 1, 2, 3];   $x = 4;   # morphs $x
<allison> it looks like the copy opcode (and removing the workaround) will solve your problem
<allison> I'll take that ticket with a high priority (unless someone else wants to grab it)
<allison> EOQ
<Coke> pmichaud: you're last in my queue.
<chromatic> If you let me know what you want coded, I can code it.
<pmichaud> thanks.  That's going to be a real showstopper until we resolve it.  I can't get 01-sanity to pass without some serious workaroundage (if it's even possible)
<allison> chromatic: okay, I'll append a quick note to the ticket
<pmichaud> designish-question:  Would there be a huge downside to having get_boolean return false on PMCNULL, instead of throwing an exception?
<pmichaud> i.e.,   in PIR    <  if $P0 then label >   throws an exception if $P0 is null
<Coke> I would tend to prefer that you have to do a null check first.
<particle> we have if_null for that reason
<kjs> there's always "if null", wouldn't that be helpful?
<pmichaud> if we do a null check first, then:
<pmichaud> (1) we have to do it for every conditional that is generated in the PAST tree
<pmichaud> (2) it's a bit of a pain because the null check is "unless" instead of "if"
<kjs> unless null also exists
<pmichaud> yes, but that means that an "if" node has to generate an "unless_null + if" combination
<pmichaud> and an "unless" node generates an "if null + unless" combination
<particle> pmichaud: is this a perl6 q, or pct?
<pmichaud> it's generated code, yes, but it just seems odd to be always generating two conditional PIR instructions every time we need a branch
<barney> alias 'if_not_null' to 'unless_null' ?
<pmichaud> particle: pct.
<particle> you can override Null with Perl6Null, which implements get_boolean and returns false
<particle> ok
<pmichaud> currently all of the languages I'm working with avoid the problem by creating 'Undef' every time a PMCNULL is detected
<allison> pmichaud: which get_boolean?
<spinclad> unless $P0 => if null $P0 or ! $P0;  if $P0 => if ! null $P0 and $P0
<pmichaud> (override Null with Perl6NULL -- works only if the aggregates return Perl6Null instead of PMCNULL)
<pmichaud> allison: get_bool for PMCNULL objects
<pmichaud> so that  "if $P0 goto label"  doesn't throw an exception if $P0 is null
<allison> ok, get_bool
<pmichaud> it's not a huge issue at the moment, so I don't need an immediate answer.  I've just come across a few cases where this would be nicer than generating the extra if_null/unless_null instructions on every conditional/loop node
<pmichaud> I just wanted to throw it out for people to think about
<Coke> Short term, you could add an experimental op that did what you wanted.
<pmichaud> I'll just generate the extra instructions in that case :-)
<Coke> I think this might be part of a larger debate about whether we should throw exceptions or return a special value for many opcodes.
<allison> PMCNULL is pretty special, I'm not fond of overloading it for dwim
<pmichaud> it wasn't as big an issue when aggregates were returning Undef for non-existent elements, but now that they return PMCNULL (as they should) it comes up a bit more often
<Coke> I am used to java, where I have to do this a matter of course.
<Coke> (not that it makes it right for parrot, mind you.)
<pmichaud> anyway, we'll take it as "no, don't change it" for now.  Thanks.
<Coke> Ok. Anyone have anything else? Anyone else blocking on anything other than time? 
<allison> I'm blocking on a chicken-n-egg scenario
<particle> chicken first, please.
<pmichaud> oh, I have a comment
<allison> I need STM, and a thread-safe interpreter, to implement threading
<pmichaud> allison -- the "hangs" that people with the concurrency changes are also showing up under Linux
<allison> and I needing threading to implement STM and a thread-safe interpreter
<pmichaud> s/people/people are seeing/
<allison> pmichaud: aye, that's my chicken
<pmichaud> several people have commented today that 'make test' seems to hang under Linux for no apparent reason.  I've had it happen twice myself.
<pmichaud> just wanted to make sure you knew it wasn't a Mac-only issue.
<chromatic> I fixed one of those hangs.
<chromatic> Not today though.
<allison> I'll turn the scheduler initialization back off
<pmichaud> I saw it this morning.
<allison> it's a plain and simple deadlock
<pmichaud> 'make test' hung, but rebuilding got past it
<allison> that appears randomly
<pmichaud> okay.
<pmichaud> I'm done.  Really this time. 
<allison> the analysis is simple, the solution, not so simple
<allison> (but possible)
<Coke> We'll need a decision regarding whether or not we'll ship 0.5.1 with that enabled.
<Coke> ... but not today.
<particle> who's the deadlock expert on p5p?
<allison> I'm debating at the moment whether to rip out the mutex/lock temporary implementation and just go ahead and get STM up to snuff
<chromatic> STM only had deadlocks on Windows, as far as I remember.
<particle> stm won't work on windows without threads
<particle> yep
<allison> option 3, is shift the temporary solution to a non-threaded solution
<particle> timeslicing, etc?
<allison> holding off on the threaded implementation until I go back and rip out the old threading implentation
<allison> particle: aye, essentially
<particle> sure. well, let's see if we can live with the performance. or, make it compile-time option
<particle> then we can test it without mucking with the souce
<particle> *source
<particle> perl Configure.pl --thread --nothread
<allison> particle: the compile-time option is a pretty simple one
<allison> it's chromatic's alternate runloop
<particle> that's runtime, no?
<particle> --runcore=...
<particle> (even better, as long as the build doesn't hang)
<allison> aye
<Coke> This sounds like we're veering into #parrot territory ... shall we adjourn there?
<particle> did pmichaud get his comment in?
<pmichaud> yes -- my comment was about linux hangs
<particle> then i'm ready to adjourn
<pmichaud> off to #parrot.  thanks everyone!
<Coke> END OF LINE, PROGRAM
<spinclad> (program loves end of line. elsewise when would it stop?)
% chromatic has left #parrotsketch
% Coke has left #parrotsketch
% jhorwitz has joined #parrotsketch
% kjs has left #parrotsketch
<jhorwitz> pasting my status for posterity:
<jhorwitz> wrote a mod_parrot HLL layer for plumhead.  can now serve up dynamic PHP pages.
<jhorwitz> after pm's fixes to the perl6 parser, got the mod_perl6 URLs working again.
<jhorwitz> beginning work on fleshing out embed.pod and friends.
<jhorwitz> EOR
% Andy has joined #parrotsketch
% Andy has left #parrotsketch
% rdice has joined #parrotsketch
% jhorwitz has left jhorwitz!~chatzilla@pool-68-236-41-146.phil.east.verizon.net
% barney has left barney!~bernhard@p549A0F1C.dip0.t-ipconnect.de
% allison has left allison!~chatzilla@dsl-241-77-141.telkomadsl.co.za
% rdice_ has joined #parrotsketch
% pmichaud has left #parrotsketch
% rdice_ has left rdice_!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com
% rdice has left rdice!~richard_d@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com
<tewk> ls