% Nicholas has joined #parrotsketch
<chip> FIRST POST
<obra> 'afternoon
<obra> fwiw, we now have a logbot
<chip> marvy
<chip> I have evil timing issues, Jesse
<obra> Oh?
<chip> I may not be present for the beginning of the meeting due to a surprise meatspace meeting.  I'll be here ASAP, and can't imagine I'll be more than 15min late
<chip> I'll be able to backscroll, though, and irssi will highlight anything starting "chip:"
<obra> Ok.
% mdiep has joined #parrotsketch
% leo has joined #parrotsketch
% allison has joined #parrotsketch
% pmichaud has joined #parrotsketch
<obra> 'afternoon, all
<pmichaud> afternoon!
<mdiep> good afternoon
<leo> hi all
<allison> hiya
<chip> hi all.  meatspace meating paused, so I'm here in the meantime
<obra> I see we have an audience today. Hi, cognominal. Because I haven't set up proper moderation just yet, I'll just ask that you hang out and watch the show, but not say anything
<obra> Also, Tweety is our logbot. Robrt got us logging set up
<obra> But it looks like we're mostly all here. 
<obra> chip, while your meeting is paused, want to give us the lowdown on your week?
<chip> My week was mild engagement on IRC, due to dayjob crunch mode.  This week I see no such problem.  I've been working on pdds; I've moved the exising pdds[*] and am recreating them in my image
<chip> [*] this broke the website mildly; I'm going to submit a patch to Robrt that should repair it
<chip> EOF
<obra> Ok.
<obra> leo?
<leo> I've been going through the DEPRECATED list
<leo> Tossed a lot of old cruft, unified interfaces, ...
<leo> mostly done, except stuff that depends on variable sized register frames (reg stack ops)
<leo> started with variable szied reg frames today
<leo> .
<Nicholas> (Hi)
<obra> Ok. Nick.
<obra> What's up?
<Nicholas> Not much. More accurately nothing. Satellite "broadband" was pathetic (real con job), some combination of IMAP server/client hates 1-3 second latency [lock the authors up on a cruise ship until they sort it out]
<Nicholas> so got little done on Perl Whirl except eating, sleeping, going to conference and having "fun"
<Nicholas> ("fun" was compulsary, and enforced regularly)
<obra> Delightful!
<obra> Ok. matt?
<mdiep> coke has been busy working with the new compiler for tcl... and we now have some inlined commands
<mdiep> although all tests aren't passing yet
% chromatic has joined #parrotsketch
<mdiep> I had a brief but productive chat with chip on irc this week about namespaces
<mdiep> now I'm reading all the threads about namespaces from this time last year
<mdiep> and I've started writing up my proposal
<mdiep> ^D
<obra> patrick?
<obra> (If folks have questions for question time, shoot them to me in /msgs?)
<pmichaud> This week I read up on Luke's L::AG stuff, and Allison's port to PGE/PIR.  I also worked with Leo on resolving the problems with coroutines/yield that was causing some PGE subrules to fail (thanks Leo)
<pmichaud> at the moment I'm figuring out how to redesign the subrule call interface to be able to better handle built-ins such as <ws>, <alpha>, <sp>, etc.
<pmichaud> I'm also considering writing versions of Text::Balanced for use in PGE rules
<pmichaud> ZZ
<chip> (back)
<allison> pmichaud: yay! on Text::Balanced
<obra> ok. allison.
<obra> what's new?
<allison> as patrick mentioned, I've been working on the PGE/PIR port of Luke's L::AG stuff
<allison> it's been a useful exercise
<allison> some ideas for the design of the compiler tools
<allison> (mainly to add a procedural interface as well as a little language for manipulating trees)
<allison> (that is, a procedural interface for defining the tree manipulation rules)
<allison> that's currently in a private svn repository
<allison> EOR
<obra> chromatic?
<chromatic> I've mostly been doing Perl 5 work lately.
<chromatic> I did check in the docs/BROKEN file, the embedded function generation code to Parrot::Vtable (without any make-it-work-magic), and a fix for a compilation error that Leo promptly fixed.
<chromatic> Er, compilation warning.
<chromatic> I'm going to look at what it takes to generate the src/embed.c file (I believe) from the Makefile and I might try once more to get rid of the JIT compilation warnings.  They run into twisty macro logic and function pointers, so they're difficult to trace sometimes.
<leo> nb: src/extend.c
<chromatic> Right, that one.
<chromatic> +++
<obra> ok.
<obra> autrijus appears to not be with us today.
<obra> So, I think that leaves me.
<obra> And since I've mostly got questions more than answers, that's a decent segue ;)
<obra> Ovid has been talking to the very kind folks at nl.net about the first two parrot milestones.
<obra> They'd very much like to hear what we/they could be doing to help accelerate parrot development.
<obra> Part of this may be a milestone rejiggering.
<obra> What could we do to M1/M2 to make it easier to tick things off?
<chip> Ah.
<chip> I think that's a hard thing to discuss on IRC
<chip> Can we go to mail for that?
<obra> Ok. Happy to take it offline.
<obra> Sure.
<mdiep> btw, it'd be nice to get something like this to show the progress on the milestones: http://trac.adiumx.com/roadmap
<obra> matt: a lot of that becomes "easy" if the milestones can get broken down into discrete tasks.
<obra> Which we're not at yet.
<obra> But sure, mail is fine.
<chip> tickets by milestone, interesting
<obra> patrick was the only one to poke me with questions.
<obra> Whcih means that now it's his floor.
<pmichaud> q1:  Leo sent out a message last week which got deleted from my inbox before I could reply to it regarding deprecation of the newsub op...
<pmichaud> pge currently uses newsub to generate its coroutines -- is there now another way to do it, or ... ?
<leo> yep the prefered way is to just clone the exiting Coroutine PMC
<leo> see also DEPRECATED
<leo> but I'll postpone the newsub stuff a bit anyway 
<pmichaud> okay.  So, any subroutine that contains a .yield() in it is automatically treated as a coroutine?
<leo> if favor of some more docss how to change what
<leo> yup
<pmichaud> interesting.
% autrijus has joined #parrotsketch
<pmichaud> okay, q2:  I think I asked this before but have forgotten the answer -- is there a simple way for a parrot sub to know if it's being called as a sub or as a method?
<autrijus> er hi.
<pmichaud> (i.e., with an invocant)
<autrijus> sorry, I was trapped in a canadian thanksgiving dinner
<obra> hello, autrijus. we're into question time. Once patrick's questions are over, I'll let you at it ;)
<chip> Hm.
<leo> while we have a bit for it in the calling conventions (unfilled) it shouldn't matter, because the invocant (or object) is passed as first argument now
<chip> pmichaud: You're hoping to say $self = called_as_method ? shift : $global_default
<pmichaud> something like that
<leo> no   o.foo(x)  :=  foo(o, x)
<chip> leo: there was going to be an invocant bit in the get_params list; is there still one?
<pmichaud> mainly, a rule can be called as a sub, or sometimes it can be called as a subrule (in which case there's a match object being used as the invocant)
<leo> < leo> while we have a bit for it in the calling conventions (unfilled)
<chip> pmichaud: is it a hardship to arrange that when you call it as a sub that you pass undef or null or something as first parameter?
<pmichaud> I'll have to think about that; initially yes, because I'm often not the one calling the sub as a sub
<pmichaud> i.e., there are other folks who are calling the sub, and it might be kinda weird to say "you always have to call the sub with a null first argument"
<leo> don't they hav a match object to pass in?
<chip> pmichaud: This is an interesting question.  There are two use cases.  There are the CGI.pm style (which I illustrated), and there are people who want to simulate a method call without actually making one
<pmichaud> no, typically when a rule is applied to a string it creates the match object request
<pmichaud> s/request//
<chip> essentialy this is an optional invocant.  ?$self anyone?  :-)
<chip> pmichaud: is it a necessary part of the design that these methods be callable as subs?
<allison> more like a self-instantiating invocant
<pmichaud> at the moment pge is looking at the type of the first argument, if it's a match object it assumes it's the invocant, if it's a string it's assumed to be the thing to be matched against and a match object is created
<leo> a :multi could help here probably
<autrijus> I think it calls for mmd.
<leo> :multi(mob, string)
<leo> :multi(string)
<pmichaud> chip: no, it's not a necessary part of the design, it's just a decision point
<leo> s/mob/Match/
<pmichaud> well, it's more typically    :multi(Match) and :multi(String)
<leo> ok, fine too
<pmichaud> i.e., if we already have a Match object, we don't need to pass the string because the Match object already holds a reference to the string to be matched
<autrijus> and the (Str) variant would invoke the (Match) variant again
<pmichaud> yeah, mmd sounds like the way to go
<pmichaud> I'll work on it from that perspective
<chip> Or a non-MMD version would be to export &MatchSubs::foo which is a wrapper around Match.new.foo
<chip> the mmd is stronger of course, just a note
<pmichaud> from another perspective
<pmichaud> ...
<pmichaud> given that I compile a rule using    p6rule("pattern")  
<pmichaud> what does it return?
<pmichaud> I mean, can it return something that is a mmd sub that understands both dispatches...?
<chip> &foo is vast; it contains multitudes
<leo> not yet probably
<autrijus> you can't return a MultiSub object?
<leo> currently an Eval object is returned that owns all compile subs
<pmichaud> I haven't looked closely at MultiSub's yet, I'm afraid :-|
<leo> I'll have a look at it
<pmichaud> anyway, I'm being prompted to move along, so we can take this to the listserv or next week if/when I have more questions
<pmichaud> EOQ
<obra> Autrijus: what do you have to report?
<autrijus> I'll keep it short
<autrijus> Pugs 6.2.10 released; we now have libpugs, in both .o and .a form, for people to link against.
<autrijus> tried to maintain compat with parrot 0.3.0 but ran into a bad assumption in parrot's embed api about get_param (and assorted other troubles); some are fixed in trunk, some I need to work out with leo afterwrads.
<autrijus> still need lex pad and var reg to target parrot; meanwhile focusing on getting object model (roles, packages, eigenclasses) to work natively and correctly
<autrijus> leo mentioned that parrot's oo dispatch can likely support stevan's metamodel primitives with reasonable efficiency.
<autrijus> which is good news.
<autrijus> otherwise, nothing much else related to parrot. EOQ
<autrijus> er, EOR.
<obra> Ok.
<obra> Who else has questions? (If not, I have one for each of you)
<chip> one per?   what a coincidence
<obra> ;)
<autrijus> also I'd like to thank chromatic for the /. submission :)
<chip> OK, that's my question: How did you figure out exactly one question per person?
<obra> chip: I cheat. To wit:
<Nicholas> [my guess is that it's the same question]
<autrijus> (although his cute choice of words seems to be a bit unfortunate :))
<chromatic> I'd like to apologize for using the word "frivolous" in a way that made no sense to anyone but me.
<autrijus> Nicholas: [ditto]
<obra> Chip, what are you blocking on?
<autrijus> chromatic: no worries.
<chip> Aha.  I am blocking on ... 
<obra> I'll batch these, so everyone can reply at once ;)
<obra> Allison, what are you blocking on?
<obra> Leo, what are you blocking on?
<allison> time, and I'm taking steps to get more of it
<obra> Matt, what are you [+tcl] blocking on?
<obra> Patrick, what are you blocking on?
<obra> Chromatic, what are you blocking on?
<obra> Autrijus, what are you blockign on?
<obra> Nick, What are you blocking on?
<chromatic> Free time and support for method attribute reflection in Pugs.
<autrijus> chromatic: which test in t/ is this?
<pmichaud> Finally, after many weeks I'm unblocked.  The merge+0.3.0 release really clears the way for me to make some progress, and allison's work on L::AG gives me good targets to shoot for :-)
<chip> obra: I'm blocking on (1) meatspace time and (2) understanding Parrot as it is.  I could make quicker work of the PDDs if I could somehow import all of Leo::Parrot::* into my brain
<chromatic> autrijus: As far as I know it's not.  I put my needs in approximate order of severity.  :)
<autrijus> I'm not blocking on anything for the 6.28.0 milestone (which would make chromatic happy). new PDDs that makes lexpad/varreg happy would be nice. a shift/reduce parser in either P6 or PIR would be nice for the next milestone.
<autrijus> chromatic: ok :)
* obra makes a note to get mind transfer tech
<chromatic> I'll take a look at the tests to see if someone added it recently.
<pmichaud> (and my other non-p6 projects are not pressing)
* leo is free running, parrot-wise
<pmichaud> autrijus:  I have my preliminary p6 parser locally, I just need to clean it up and document it for check-in
<mdiep> I'm blocking on free time
<pmichaud> er, pir shift/reduce parser, I mean
<autrijus> pmichaud: I'm not blocking on the cleaning up and documentation, fyi :)
<pmichaud> autrijus: roger that
<chip> status += I've contemplated having Leo put his new work on a branch, but I've decided instead to monitor what he's doing and be ready to cry foul in realer time
<Nicholas> contiguous free time where I can sit down and think without having to task switch back to other stuff
<chip> This is part of leo non-blockage, thus the note
<obra> Cool. from now on, I'm going to ask that question earlier ;)
<obra> Anything else I can do for folks to help push things along?
<autrijus> obra: make that the first question of the meeting sounds like ~good idea
<chromatic> I thought the question might be "What is a mailing address so lurkers can send you presents and well-wishes?"
<chromatic> I think a lot of silly things sometimes.
* chip squelches about a half dozen goofy questions
<obra> Right then. 5 minutes. anything else before I let chip unsquelch the goofy questions?
<autrijus> hm, I'm looking for ideas for a euro oscon lightning talk; the deadline for submission is in 3 hours, anyone has great ideas?
<Nicholas> actually, s/contigous free time/contiguous time/, given that ponie is work sponsored and free time project is 5.8.x
<chromatic> autrijus: fix a failing test in your lightning talk.  Pick a random one.
<chromatic> Just show people how you code.
<pmichaud> chromatic++
<obra> autrijus: how about a quick rundown of cool new language technologies none of us have ever seen?
<chip> autrijus: show why svk
<autrijus> chip: clkao has a euroscon talk wrt svk
<chip> ok
<autrijus> allison: want to do random poetry performance thing with me?
<autrijus> anyway, I'll think about the ideas :)
<autrijus> chromatic: I don't think people can follow coding-on-projector well. I can't myself
<allison> autrijus: sure :)
<chromatic> autrijus, I don't think people can follow coding with you period.  I think it's good theater though.
<autrijus> obra: I guess I can have a talk about how visual basic rocks.
<autrijus> allison: ooh :)
<obra> Right then. See y'all next week.
<chip> thanks obra
<autrijus> okay, I'll submit a "VB rocks" one and figure out something with allison.
<mdiep> visual basic? *shudder*
<pmichaud> next week, thanks as always obra
<leo> thanks obra 
<autrijus> obra++
<autrijus> mdiep: type inference, DSL macros, query comprehension and combinators, ahead-of-time compilation, optional dynamic/staticness, and more!
* chip hooks up a generator to Dijkstra's spinning corpse
<autrijus> (STM too, apparently.)
<mdiep> autrijus; I'm finishing up a $work project in VB6. it doesn't have continue/next or break/last (and the IDE works sporadically)
<autrijus> mdiep: ah, but this is VB9.
% pmichaud has left pmichaud!~chatzilla@c-24-1-26-255.hsd1.tx.comcast.net
<mdiep> ah
<autrijus> mdiep: the VB team has been thoroughly infiltrated by Haskell heads between VB6 and VB9 :)
<Nicholas> autrijus: what do Candians give thanks for?
<autrijus> Nicholas: turkeys, I hear.
<mdiep> autrijus: that sounds promising :-)
<autrijus> one of them said "turkeys are so tasty, if they are more intelligent they will eat themselves"
<Nicholas> this seems unlikely
<chromatic> They don't know how to cook.
* autrijus decides to detach this particular thread
<Nicholas>  abuse@public1.sz.js.cn
<Nicholas> no, that's not what I was thinking
<chromatic> Do tell.
<Nicholas> well, it had a lot to do with a whois lookup on 218.4.90.19, which in turn is less interesting than turkey
<Nicholas> Time seemed to be a common theme in the answer to Obra's question
<autrijus> chromatic: Emit.PIR.ParrotObject in pugs tree is what I was talking about last week.
<chromatic> Ah, thanks autrijus.
<autrijus> chromatic: it is _PIR_ tree not _PIL_ though, but can easily be adapted to do PIL.
<chromatic> I'm writing that down to explore, after I do Parrot::Test.
<autrijus> chromatic: but still, if I know how you plan to visit the tree, I can probably emit them in more useful form.
<obra> right. y'all should move back to a non-meeting channel, like #parrot :P
% leo has left #parrotsketch
% chromatic has left chromatic!~chromatic@sub17-30.member.dsl-only.net
<autrijus> thanks again, obra :)
% autrijus has left #parrotsketch
% mdiep has left #parrotsketch
% allison has left allison!~allison@sub17-30.member.dsl-only.net