% contingencyplan has left contingencyplan!~contingen@76.186.27.146
% rdice has joined #parrotsketch
% rdice has left rdice!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com
% rdice has joined #parrotsketch
% pancake has joined #parrotsketch
% pancake has left pancake!~foo@acpl.adam.es
% pmichaud has joined #parrotsketch
% parrot-poke has joined #parrotsketch
% parrot-poke has left #parrotsketch
% allison has joined #parrotsketch
% chromatic has joined #parrotsketch
% barney has joined #parrotsketch
% Coke has joined #parrotsketch
<Coke> I am here today; if I go idle for too long, I'm sure chromatic will jump in as before.
<chromatic> Yep.
<Coke> very good. Why don't we start with allison, barney & chromatic, in that order.
<particle> hi-o
<allison> - I've been working on events, which means working on the concurrency scheduler
<allison> - The interpreter structure has a PMC* element, to hold the concurrency scheduler
<allison> - Interpreter has initialization code to set up concurrency scheduler
<allison> - Scheduler runloop runs as a separate thread (so parrot now runs 4 threads  every time it starts up instead of 3, but that will drop back down to 2 once the migration is complete and the 'event' and 'io' threads are merged into the concurrency scheduler)
<allison> - Per the coding standards, C routines for the concurrency scheduler API contain the prefix 'CX' for Concurrency Xcheduler ('CS' is far too heavily used for other things).
<allison> - I've checked in the files for the concurrency scheduler runloop, but the pieces that initialize and destroy the concurrency scheduler are currently commented out, while I finish off some final bits of runloop termination. (Search for 'Parrot_cx' in src/inter_create.c to uncomment and try it.)
<allison> (actually, that was pre-typed, and I'm checking it in now)
<allison> EOR
<barney> Ported Plumhead over to PCT and new PAST.
<barney> .eor
<chromatic> I did a bit of work on the concurrency/events PMCs.
<chromatic> I did some minor cleanup.
<chromatic> It looks like I'll have a few hours on Friday night to hack on things in an airport, so if anyone wants to send me some small, bite-sized tasks to work on, I should be able to get to a few of those.  It's not likely I'll have much time before then though.
<chromatic> EOR
<Coke> Here's a report from kid51:
<Coke> 1.  Refactored Parrot::Configure::Data and config step classes so that the Perl
<Coke> 5 %Config and special variables are queried for settings to be added to the
<Coke> Parrot::Configure object only in step #2, init::defaults.
<Coke> 2.  Created a '--silent' option to Configure.pl.
<Coke> 3.  In 'capture' branch, began process of replacing use of
<Coke> Parrot::IO::Capture::Mini with non-core CPAN module IO::CaptureOutput.  Results
<Coke> so far are promising.  If fully successful, will recommend that this module be
<Coke> included in the trunk distro.
<Coke> 4.  Am considering various approaches to revising the testing of the
<Coke> configuration steps to respond to the failing cases reported by Andy Dougherty
<Coke> and chromatic.  Began working on this in the 'nointer' branch, but approach I
<Coke> was taking there will probably be scrapped.  Two other branches, 'archjit' and
<Coke> 'cpu', are temporarily stalled while I ponder these issues.
<Coke> 5.  Am trying to delegate parts of #3 and POD revisions to new project members.
<Coke> 6.  Solicited comment on a variety of older tickets which are still open. 
<Coke> Where I did get feedback, resolved some tickets.
<Coke> 7.  Would love to have an overhaul of our smoke reports page so that (a) I
<Coke> EOR
<Coke> kjs:
<Coke> + A bit of work on PIRC's back-end, creating a parse tree.
<Coke> + created a compiler driver, which invokes each of the three compilation
<Coke> phases (heredoc, macro, PIR). This makes using PIRC easier. It now also
<Coke> allows for pre-process only options. (No Makefile yet, so compiling PIRC
<Coke> must be done manually (not hard, but annoying))
<Coke> + Trying to link to IMCC's C<is_op(char *name)> so that PIRC can parse all
<Coke> Parrot ops; didn't work out, due to #include hell.
<Coke> + Reading a bit on register allocation theory.
<Coke> EOR
<Coke> me:
<Coke> - Continuing work to manage/organize RT queue
<Coke> RT - http://www.perlfoundation.org/parrot/index.cgi?useful_rt_queries
<Coke> - Deprecation hunt; trying to eliminate all items marked for deprecation
<Coke> post-0.5.0; done a few of the simple ones; trying to ID issues involved
<Coke> with deprecating certain items. q< use more tickets >; Next step, go
<Coke> through docs/sources and insure that all items marked as deprecated get
<Coke> a ticket, even if we don't know what release they're getting deleted in.
<Coke> As part of pre-removal sanitizing, reduce usage of deprecated items making
<Coke> it easier for the actual removal to go smoothly.
<Coke> - Trying to track down issues in tcl on feather; Could use someone to
<Coke> help investigate RT #46973 (linking issues on osx/86 and osx/ppc).
<Coke> - Someone on #parrot mentioned the python tutorial, and how having a
<Coke> cohesive tutorial for PIR would be helpful. I think a PIR tutorial
<Coke> would be an excellent idea; Probably best to keep it in the wiki for
<Coke> now to simplify contribution. If I don't hear any objections, I'll
<Coke> get that setup shortly. Feedback on seeding with content from the
<Coke> ... missed some content there. 
<Coke> repository and copyright issues welcome. (future tutorials could
<Coke> include other areas, but I think concentrating on PIR to start with
<Coke> is a good idea.)
<Coke> - Gave my first monthly report regarding milestones, hopefully freeing up
<Coke> Allison to write more code. =-)
<Coke> EOR
<Coke> There was a bit chopped off kid51's, too:
<Coke> 7.  Would love to have an overhaul of our smoke reports page so that (a) I
<Coke> could easily get email addresses for testers who are reporting failures in
<Coke> tests I wrote; (b) I could search all results by test and thereby easily
<Coke> compare results across platforms; (c) sort by OS/platform; (d) sort by date of
<Coke> reports (some year-old reports appear at top of page).  Would this be a good
<Coke> subject for a microgrant?
<Coke> EOR>
<Coke> particle, perljam, pmichaud/
<particle> it's been a productive week, which feels GREAT!
<particle> pmc:
<particle> ~ added get_number vtable method to FixedPMCArray,
<particle>   should likely also be added to all Fixed*Arrays (cage task?)
<particle> pct:
<particle> perl6:
<particle> ~ converted some pir to nqp
<particle> ~ got most of 00-parrot and many 01-sanity tests passing with nqp-based perl6
<particle> ~ modified if/unless to match latest spec
<particle> kp6:
<particle> ~ spoke to kp6 team members about kp6-on-nqp
<particle>   --hoping to help fglock get started on a port this week
<particle> .end
<pmichaud> ENOPERLJAM
<Coke> not hearing PJ, PM?
<pmichaud> I'm still writing up my report -- just got back frome errands about 10 mins ago.  Come back to me
<Coke> lesseee... spinclad? tewk?
<particle> coke: please send a listmail about the pir tutorial task, when the infrastructure is set up. i have a feeling that advertising it could help a bunch
<Coke> Sure.
<Coke> let's move into questions for a bit.
<pmichaud> okay, I have my report ready
<pmichaud> ...reporting:
<Coke> ... ok, let's rollback our question transaction.
<pmichaud> * More pct items completed: repeat_until, try, chain, copy, xor
<pmichaud> * Refactored PCT so different components may be easily loaded separately
<pmichaud> * More improvements to nqp
<pmichaud> ** particle++ and others helped prototype an inline pir capability
<pmichaud> ** including a heredoc version
<pmichaud> * Completed implementation of 'abc' in nqp
<pmichaud> * Added the blue_rect.pl example to examples/sdl/
<pmichaud> * Added lots of explanatory comments in abc, and some in nqp also
<pmichaud> * Removed a lot of obsolete branches from the repo
<pmichaud> * Added a 'make coretests' target, and refactored the test harness to clean things up
<pmichaud> * Updated PGE syntax
<pmichaud> ======
<pmichaud> * Working on the perl6 rewrite in NQP
<pmichaud> * It is going _very_ quickly
<pmichaud> * I expect to merge back to trunk at the conclusion of #ps
<pmichaud> * The quote functions are _much_ nicer in nqp than in PIR
<pmichaud> end of report
<Coke> danke.
<Coke> I have some questions. Allison, for you, from my report; any issues you see with copyright if we move some stuff from the repo into the wiki for documentation cleansing and then back into the repo?
<Coke> does TPF care?
<allison> the wiki is also TPF owned
<allison> no problems
<Coke> content too? excellent.
<Coke> particle, go.
<particle> allison: is there a single-threaded execution model for parrot, or is two threads an absolute minimum in the new world order?
<allison> at the moment (before my changes) 3 is the absolute minimum
<allison> there is a single-threaded variant spec'd in the concurrency pdd
<particle> great
<allison> not all platforms support threading
<allison> but, it's likely that the threaded version will be preferable, in general
<particle> yes, that was my concern. glad it's been addressed. i look forward to properly reviewing the pdd soon
<Coke> eoq?
<particle> eoq.
<allison> because the non-threaded version will have to pause execution of ordinary code to finish io, events etc
<allison> eoa
<chromatic> Seems like a variant runloop could handle that.
<chromatic> It's how single-threaded Perl 5 handles async signals, after all.
<allison> yup
<chromatic> Good.  I basically have that written then.
<allison> currently parrot does some nastiness with swapping out function pointers
<allison> but that means events don't work at all in a JITed environment
<particle> metronome is a java gc whose arch could be studied for ideas on reducing runtime impact
<allison> particle: hmmm... I'll take a look
<allison> chromatic: what, a variant runloop for parrot? ah, yes, the gc one you were working on
<chromatic> Same principle.
<allison> yeah, we can just slot in "check tasks" instead of 'run gc'
<allison> (and make it as cheap as possible to check tasks)
<particle> metronome is hard realtime gc, and looks lovely. i'm sure principles could be applied to concurrency, perhaps the infra could even be shared across concurrency and gc
<chromatic> Same principle between those, too.  "Do I have a consistent view of memory?"
<allison> aye, it's the "real-time scheduling" that's key
<Coke> ok. let's close that up and move on to patrick's questions...
<pmichaud> I have quite a few
<pmichaud> simple one first:   t/oo is not currently part of 'make test'... should it be?
<Coke> you have the floor. let me know when you're done. =-)
<Coke> pmichaud: yes.
<pmichaud> followup to that:  can anyone work to get t/oo passing, first ?
<allison> I'll get it passing
<allison> (surprised it's not)
<pmichaud> allison:  excellent.  This is ticket #47924
<pmichaud> apparently t/oo was never added to 'make test'
<pmichaud> next item isn't a question but more of a 'heads up'
<pmichaud> at some point I think we will need to take a good luck at lexicals in parrot again.
<pmichaud> s/luck/look/
<pmichaud> currently :init/:load subs cannot be :outer for other subs, and methods cannot have the :outer flag on them (i.e., they cannot be lexically embedded in other blocks)
% guru has joined #parrotsketch
<pmichaud> I'm putting a variety of workarounds in PCT for stuff like this, but eventually language features will run into Parrot limitations and we'll have to solve it
<pmichaud> any comments on lexicals before I move on?
<pmichaud> I could also use a good explanation of exactly what the 'newclosure' op is, what it does, and when it should be used (or shouldn't)
<allison> pmichaud: on t/oo, the failing tests are passing now
<pmichaud> allison++
<pmichaud> okay, nothing on lexicals then.  moving on...
<pmichaud> any idea when we might have the new <die> opcode implemented from pdd23?
<pmichaud> there are a few times when I would've liked to have used it
<allison> pmichaud: 'newclosure' creates a new Closure PMC, capturing context
<allison> useful because it throws a fatal exception?
<pmichaud> useful because I can pass it a string instead of creating an Exception object first
<allison> there are a few things about exceptions that made more sense to wait on until the concurrency pdd was implemented
<pmichaud> okay, that's a good enough answer for me :-)
<allison> that one could probably have the interface set up now, and just filled in with the right implementation later
<pmichaud> I think I'll just send my remaining stuff to the list
<Coke> ... ok.
<Coke> Anyone have anything that they are trying to get into 0.5.1? Add the ticket to the child-list of the release ticket for me.
<Coke> (We haven't had good luck with that in th epast because we added too much. if we just use it to reflect reality rather than desire, I think it'll be more useful.)
<Coke> Anyone else have any questions?
<Coke> (also, the usual: anyone not having fun? blocking on something other than time?)
<pmichaud> I have some specific features that are blocking on other Parrot features, but other than that I'm blocking on time
<chromatic> Is one of those the ccache weirdness?
<pmichaud> no, I'm not blocking on that (thankfully)
<pmichaud> I just reported it because it was causing 'make test' to fail on my machine
<chromatic> Okay.  I haven't made progress on that.
<pmichaud> it's not in my critical path at all
<pmichaud> once I was able to confirm that it wasn't a change I made that caused the problem, I filed the ticket and forgot all about it :-)
<pmichaud> oh, I have another question
<Coke> go
<pmichaud> what's a good way for me to indicate what rt tickets I would find most useful/beneficial for the perl6 work?
<chromatic> Meta-ticket?
<pmichaud> i.e., to prioritize them
<pmichaud> I could do periodic items to mailing list, notes in journal postings, wiki page, meta ticket, reports here....
<allison> I think a wiki page makes the most sense, with just a link to the relevant tickets
<Coke> pmichaud: for now, co-opt the useful_rt_queries page on the wiki?
<allison> (that way we can constantly update the page as tickets are closed)
<pmichaud> well, I could also make a query in rt that does what I want, and use the rt priority attribute
<barney> Meta tickets are nice, becaust the status is always up to date
<Coke> barney: one issue is that resovled tickets tend to hang around and clutter the list; but yah, that works.
<Coke> pmichaud: for now, if you can just get the list together...
<barney> Coke: OTOH there is a sense of achievement
<Coke> what!? and boost morale!?
<pmichaud> I'll put it as a wiki page for now, and we can migrate it somewhere else if it's not being effective there.  thanks.
<Coke> sounds good.
<Coke> Anyone else?
<particle> who's on release duty this month?
<Coke> The list says allison. We're going to try to swap smash in, if he's still amenable.
<Coke> And, while I was going to wait a bit; we could use volunteers to be release managers in 2008.
<particle> that's as i recall. okay, we need to check in with smash and make sure he's still on, then get him up to speed
<Coke> (if you're already a volunteer, just check in with me to make sure you still are.)
% jhorwitz has joined #parrotsketch
<particle> since we'll have a new release manager, let's make sure we have a mess of bugs and problems for him to sort out
<pmichaud> I've been filing them quickly
<Coke> jhorwitz: we're just wrapping up here. Anything you want to slip into the minutes?
% schmalbe has joined #parrotsketch
<jhorwitz> been complete idle these past two months.
<jhorwitz> presenting mod_perl6 at frozen perl so i'm hoping to get back up to speed ASAP
<pmichaud> jhorwitz: have you looked at nqp yet?
<jhorwitz> i've followed the various threads, but haven't poked around yet.
<pmichaud> ah.  It has classes and methods.  :-)
<jhorwitz> looks exciting!
<jhorwitz> pmichaud++
<jhorwitz> :)
<Coke> I have an obligation; Good week, everyone. chromatic, if you can bring us home...
<jhorwitz> i had nothing else.  EOR
<particle> i'll fill in, coke.
<particle> see you all next week!
<allison> thanks all!
% allison has left #parrotsketch
% pmichaud has left #parrotsketch
% barney has left barney!~bernhard@p549A0881.dip0.t-ipconnect.de
% guru has left guru!~guru@TOROON01-1168098147.sdsl.bell.ca
% guru has joined #parrotsketch
% Coke has left #parrotsketch
% contingencyplan has joined #parrotsketch
% chromatic has left #parrotsketch
% schmalbe has left schmalbe!~bernhard@p549A0881.dip0.t-ipconnect.de
% guru has left guru!~guru@TOROON01-1168098156.sdsl.bell.ca
% guru has joined #parrotsketch
% guru has left guru!~guru@TOROON01-1168098147.sdsl.bell.ca
% rdice has left rdice!~richard_d@CPE0014bfafbbd5-CM0011e6ecf48a.cpe.net.cable.rogers.com
% jhorwitz has left jhorwitz!~chatzilla@pool-68-163-17-123.phlapa.east.verizon.net
% rdice has joined #parrotsketch
% rdice has left rdice!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com
% rdice has joined #parrotsketch
% rdice has left rdice!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com
% rdice has joined #parrotsketch
% rdice has left rdice!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com
<spinclad> (late response:) ENOREPORT (& EINBEDAILINGFEELBETTERNOWTHX)