% Coke has joined #parrotsketch
% leo_ has joined #parrotsketch
% Psyche has joined #parrotsketch
% chromatic has joined #parrotsketch
<obra> hi gang
<particle> hi-ho
% pmichaud has joined #parrotsketch
% allison has joined #parrotsketch
<chip> howdy y'all
<obra> hey chip
<leo_> hi all
<obra> ok. I suspect audrey's asleep, as she was hacking until far too recently.
<obra> let's dive in.
<obra> I hope everybody brought their pre-typed status reports ;)
<obra> So. Chip, what's new?
<pmichaud> (audrey appears awake on #parrot... but perhaps that's just a minimized window)
% audreyt has joined #parrotsketch
<chip> ah, paste time
* chip hopes he doesn't get kicked
<chip> Past week:
<chip>  * Having a lot of fun coding & coordinating
<chip>  * Implemented the first phase of PDD21 Reloaded:
<chip>      get*namespace opcodes, Namespace methods,
<chip>      Parrot_get_namespace*() C functions
<chip>  * Added vsoni and tewk as committers; helped vsoni with some ideas
<chip>    (leo++ has also been helping vsoni, thanks leo)
<chip>  * Finally gave leo a couple of ticket #s to work on :-)
<chip>  * Sketched some code cleanup ideas for the cage cleaners
<chip>    (andy++ has been collating these into cage/todo.pod)
<chip> Future week:
<chip>  * Implementing the rest of PDD21 Reloaded:
<chip>      get*global and set*global opcodes
<chip>  * Phasing out obsolete namespace APIs
<chip>      (find_global ops, Parrot_find_global* C functions)
<chip>  * Release 0.4.7
<chip>  * Getting ready for OSCON - updating Parrot presentation
<chip>  * More fun
<chip> Infinity and Beyond:
<chip>  * Untyping the default namespace
<chip>      this will require a powwow, likely in Portland
<chip>  * PDD23 Reloaded (Exceptions)
<chip>  * Even more fun
<chip> //
<obra> (To not embarass anyone who's frantically typing, who's ready to go next?)
% woggle has joined #parrotsketch
<Coke> me.
<obra> go for it, coke
<Coke> o been herding some RT tickets.
<Coke> o working on converting *part* of tcl ([expr]) to use PGE/TGE.
<Coke>   Apparently I should have been paying more attention when pmichaud   set me up on APL. I am at the point now where [expr 1] gets through
<Coke>   PGE, the first TGE pass... and then goes into an infinite loop on the
<Coke>   second. Should have something by next parrotsketch.
<obra> (EOR?)
<Coke> EOR. Yes. (thought the lack of new lines being pasted was obvious, sorry.)
<obra> heh
<obra> chromatic: what's exciting?
<chromatic> I fixed one segfault in CLI argument handling and apparently found a few more.
<chromatic> I updated the Perl 5 and Ruby bindings with the namespace changes.
<chromatic> I'm close to checking in the Perl 5 bindings to the Parrot repository.
<chromatic> I think I know how to write a simple IO layer to handle the STDOUT/STDERR redirection I want for the Parrot::Test revision.
<chromatic> That's all.
<obra> leo_: you up for it?
<leo_> * tested & applied vsoni's imcc exception handler patch added failing eval.t - needs still work
<leo_> * tested vsoni's register_move patch several times 
<leo_>   applied today after a lot of conversation
<leo_> * fixed the socket / accept bug - boy this code isn't really nice
<leo_>   (struct sockaddr packed into a STRING* and what not ...)
<leo_>   also fixed httpd.pir
<leo_> * fixed and closed old ticket WRT <cmp>_p_i_ic opcodes
<leo_>   these did use get_integer, which was plain wrong, e.g. for BigInt lhs,
<leo_>   or Float (the tests were also wrong for the latter)
<leo_>   compare ops with native NUM and STRING needs also some work WRT MMD
<leo_>   (these are using get_number and get_string currently, so they don't MMD)
<leo_> * we really need pddXXscalar-*.pod  
<leo_>   all works handwavingly as is, except in cases, when it doesn't
<leo_> EOR
<obra> allison: how about you?
<allison> Design discussions on and off list, mainly about namespaces, the:immediate pragma, a .loadlib directive, upgrading/replacing IMCC, and starting an OST-to-bytecode compiler.
<allison> Wrapped up a revision of the namespaces PDD.
<allison> Looking forward, my first priority is still zipping through existing PDDs, but will tackle other PDDs after that.
<allison> EOR
<obra> pmichaud: how's being back to hacking?
<pmichaud> This week I've worked on building a standard compiler base object for the various language compilers
<pmichaud> Little to report on that in terms of code; still working out some of the desirable interfaces
<pmichaud> Submitted some suggestions for pdd21
<pmichaud> Otherwise I've been following the discussions about parrot, pdd21, pge in C, and the like
<pmichaud> And getting local affairs in order for my trip to OSCON next week
<pmichaud> EOR
<obra> particle: you aruond today?
<particle> sub report
<particle>   say "playing with parsers: Perl6, TAP, bc, and now (today) PIR" 
<particle>   say "waiting on a commit from patrick for parrot compiler tools, then i'll convert these to use those tools." 
<particle>   say "adding tickets for cage cleaning activities and assisting with the learning process for the new cleaners" 
<particle>   say "wondering about the sunday hackathon... i'll reserve that for question time." 
<particle> .end 
<chromatic> s/sub/\.sub/
<obra> heh
<particle> paste-o.
<obra> audreyt: you're up
<audreyt> Still working on IMCC reentrancy.  With vsoni++'s newly commited code, Pugs can now embed Parrot without occasionally dying unexplicably.  Probably not going to change compilers/ast/ for reentrancy -- would like clarification on its status in the roadmap.
<audreyt> Also, looked into how Manhattan MMD can be retrofixed, but no code yet.
<audreyt> EOR.
<obra> woggle: how's the hacking?
<woggle> Got dynpmc and dynoplibs to work in spawned threads.
<leo_> woot
<chip> w00t indeed
<woggle> Did some performance hacking on STM, yielding substantial improvements.
<obra> (anything else?)
<woggle> In the process, I discovered that Parrot_clone() is apparently much slower than VTABLE_clone() for things like Integer. I'm guessing similar holds true for Strings and such. So it would be nice if I could avoid using it for cloning values into new interpreters while getting a deep, non-COW shared clone.
<woggle> I've been kind-of blocking on exception issues w.r.t. runtime support for STM, but I think I'll try to proceed in that regardless.
<woggle> EOR
<obra> And me, I'm just trying to put together an accurate state of the perl6 / parrot world and roadmap that doesn't overpromise, sell ourselves short or frustrate any of the core hackers
<obra> And with that, we're into question time.
<obra> Chromatic first, then allison to answer a question of audrey's, then chip, then whomever pings me
<obra> (also, chip++ for pushing us to pasted reports)
<chromatic> Okay, what was my question again?  I don't think I had one.
<pmichaud> particle had the question
<obra> oop.
<obra> particle, then.
<pmichaud> (regarding hackathon plans)
<obra> Last I heard, folks were going to try to find space on sunday to hack. The doubletree was suggested. I'm staying there, so I'll see if I can find a sane lounge.
<pmichaud> I vote for doubletree, at least as a meeting point.  If not, then I'll be in the Red Lion and can be reached there
<audreyt> seconded (I'm in doubletree too)
<particle> i'm driving in from seattle, and wo'nt have oscon credentials. just making sure i can get to the hackathon.
<obra> particle: that shouldn't be a problem
<particle> excellent!
<obra> from monday, we get free and open space at oscon.
<chip> I'm walking distance to the doubletree.  We'll need networking.  Should we bring an AP?
<obra> if you've got one handy, that'd be great.
<chip> I think I do.  I'll mail to p2 when I know.
<obra> ok. anything else on this?
<audreyt> "p2" = "p-p"?
<obra> parrot-porters, I believe
<obra> allison, you're up.
<chip> p2 = p^2 = pp, yeah
<allison> audrey: we decided not to use pp for the abbreviation because it causes snicker when said aloud
<allison> which first, audrey's question?
<audreyt> well... "p-dash-p" is okay I guess, but sure I'll adjust
<obra> allison: go ahead and do both yours. 
<Coke> why not p1p. =-)
<obra> you had an answer for an audrey question and one of your own, yes?
<allison> Status of compilers/ast: it was a prototype
<obra> Coke: because that list is "allison and schwern" ;)
<audreyt> Coke: that's perl1-porters
<allison> not currently maintained, and not in the future plans
<Coke> parrot1parrots. it's not my fault all these projects start with pee.
<pmichaud> would it be reasonable to rm compilers/ast ?
<Coke> er, "parrot1porters". curses.
<audreyt> or at least remove it from the build process and reflect this in README?
<allison> pmichaud: I'm not sure what is depending on it at the moment, but long-term, yes, it's removable
<audreyt> compiler/ast/README that is
<allison> go ahead and update the README
<allison> verify dependencies before removing from the build process
<allison> if you find any, run them by the list so we can decide how important they are
<audreyt> sure
<Coke> (don't forget DEPRECATED)
<allison> okay, my question ...
<allison> PGE in C: does anyone know who's idea it was? Are people excited to work on it? Would people rather not work on it? Is making the current PGE able ot output C compilers (a la bison, et al) good enough?
<audreyt> pmichaud started PGE in C, so it's certainly originally his idea :)
<obra> I first (recently) heard it as "Patrick's idea" from a third party.
<Coke> in general my take is 'features first, speed second'. 
<obra> But I'm not convinced that Patrick was the driver behind the recent interest
<allison> aye, (and my idea) but he rejected it because Parrot had so many features to offer
<pmichaud> I was going for features first
<chip> I think I came up with it after thinking about bringing PGE to the masses.
<allison> (and that was years ago)
<chip> Sort of like libp6re
<pmichaud> I started out in a libp6re direction, but not having unicode available (and coroutines) was looking a bit harried
<chip> Indeed, it was an idea, not a request, and C is a wasteland of lack-of-features.
<audreyt> I wonder, aside from a fast dynamically resizable array and hash, as well as unicode strings and lack of coro, is there anything else?
<chromatic> Lack of compelling use case?
<pmichaud> audreyt: those are the biggies.  Plus having a convenient mechanism to interface it with Parrot
<chip> audreyt: exception handling, host-environment namespace hooks ... but now we're just talking implementation rahter than environment
<audreyt> err, exception handling couldn't be a showstopper, could it? PGE had done without :)
<chip> oh, and GC is a big one ... if you can't count on that you're back to refcounting
<chip> s/big/irritating/ perhaps
<audreyt> so, there's C-land things that address them: judy, re2c, boehm, etc.
<allison> The thing is, if we can output reasonable C compilers from the current PGE, there's much less motivation to actually reimplement it.
<audreyt> s/re2c/c2lib/
<Coke> yes, but then parrot has *two* sets of each of those to maintain.
<chromatic> Lack of a freakin' compelling use case?
<audreyt> chromatic: would you not like writing rules in javascript?
<allison> Parrot becomes a maintaner dependency, rather than a user dependency.
<audreyt> chromatic: or in your Ruby code?
<chromatic> No.  I think that sucks.
<chromatic> I want Perl 6.
<audreyt> ok. then it's compelling to me but not you :)
<chromatic> I don't want JavaScript or Ruby or Pheme or PHP.
<Coke> (he does want tcl, though. he told me so!)
<chip> Coke: not if you do it right.  libp6re/configure --with-parrot implies using Parrot facilities for all that stuff
<audreyt> chromatic: wait, libp6re sucks more than libpcre?
<pmichaud> allison:  is it speed or portability that is driving the question...?
<chromatic> For the purposes of getting Perl 6?  Yes.
* chip takes no sides on the "should" question, but finds the "how" questions fascinating
<allison> For me it's a question of developer resources. Is this the best way to spend our time right now?
<pmichaud> I've been working from the assumption that it's not the best use
<obra> I don't think it makes sense to divert resources to p6re.
<audreyt> allison: so, emitting C has two flavours: one that allows runtime modification of grammar, one without
<obra> However, if some talented outsider were to show up and hack on it, that would be cool.
<audreyt> allison: I think the "without" one is much smaller in scale and in practice sufficient.
<pmichaud> however, coming up with a C emitter for PGE would not be difficult and could be very worthwhile
<leo_> why emitting C - put that effort in making Parrot faster ;-)
<allison> I'm fine with saying that C emitting never allows runtime modifications of the grammar
<obra> audreyt: or do you feel that not having p6re is going to screw us in the medium term?
<pmichaud> speaking for me personally, I'd rather focus efforts on improving the state of the compiler tools suite than re-writing pge at this point
<allison> patrick: yes, I think that makes sense
<pmichaud> I'll also note that if we get parrot working well, we get p6re capabilities in JavaScript, Pheme, PHP, Ruby, etc. directly
<pmichaud> even if PGE remains parrot-based
<chromatic> I already have p6re capabilities in Ruby and Perl 5 directly.
<obra> Given the fact that this is largely a volutnteer effort, I don't think anybody can make someone focus on what they don't want to.
<pmichaud> (and assuming we're able to get those other languages to target parrot, which is also being worked on and consistent with the original plans for parrot)
<audreyt> obra: a compehensive spec-based p6rule test can achieve the same medium-term goals as libp6re
<audreyt> which is consistency across p5land and parrotland and otherland rule implemeentations
<audreyt> pmichaud: is that in the tree yet? sorry if I missed the log
<obra> audreyt: ok. (and does anybody think that a rules test suite is a bad idea?)
<Coke> no.
<obra> (if so, come here and I'll knock you around a bit ;)
<pmichaud> audreyt: it's not in yet -- I got distracted onto the compiler bit
<chromatic> Good idea, but I still don't care about Perl 5 land and otherland.
<audreyt> chromatic: I do. deeply.
<pmichaud> let me reprioritize myself a bit, to get the p6rule suite into parrot
<audreyt> pmichaud++
% obra is now known as moderator
<audreyt> with that, libp6re is not blocking for anything medium term
<chip> moderator++
<pmichaud> it won't take long -- I just need to carve out some time.  There are some things that don't spec well just yet
<moderator> I think it's probably wise not to focus on "what I wouldn't find useful" unless there's a compelling reason another interested party working on it would be detrimental to your work.
% moderator is now known as obra
* pmichaud parses "not", "wouldn't", "unless", "detrimental"
<audreyt> quad negatives!
% obra is now known as moderator
<chromatic> Basically, obra just said "No negative ninnies, chromatic."
<moderator> Don't bitch about things other people want to do, unless it screws everyone
% moderator is now known as obra
<pmichaud> merci, moderator
<obra> chromatic: no, just come up with a better argument than "I don't see the point"
<obra> ;)
<obra> and with that
<chromatic> It's a distraction.
<chromatic> We're spending too much time debating NIHs.
<obra> It's a distraction for people who will be distracted anyway.
<pmichaud> "Don't distract from Perl 6" is a valid point, IMO.  But I agree we don't stop others from progressing 
* chip negates negativity and disappears into a vortex of self-reference
<obra> audrey isn't about to get patrick to drop parrot ;)
* allison throws up an interrupt flag
<obra> anyway
<chromatic> I'm sick of reading debates in RT tickets.
<allison> This was my question and I can end it.
<audreyt> allison++ # timely destruction
% chromatic has left chromatic!~chromatic@sub17-30.member.dsl-only.net
<obra> chip had a question
<chip> woggle, did you survive the last question?
<woggle> Yes.
<chip> Excellent.  I'm interested in the posssibility of merging STM before it's feature-complete for two reasons
<chip> One for us, so we can have working threads again
<chip> Two for you, so that your trunk-merges are less painful.  Esp. since there are code cleanups that could be semantically lightweight but textually huge
<chip> What think?
<woggle> Sounds like a good idea.
<chip> Would the test status degrade -- are there new failing tests if you merge?
<Coke> how does that effect the language people?
<woggle> Yes.
<chip> Hrm.  Anything you or we can do about that except "merge later"?
<woggle> I could TODO them, I guess.
<particle> let's TODO and make them a priority for the next release
<chip> let me rephrase
<particle> (or whenever if they're STM-specific tests)
<woggle> One of them should be 'easy'.
<woggle> The others are STM-specific.
<chip> are any of the _existing_ test that now pass going to fail, or are the newly-failing tests new on your branch?
<chip> s/any/many/
<woggle> I think I don't have any newly failing tests, except for t/pmc/threads.t test #1 which is skipped in trunk.
<leo_> chip: what's the time frame for next release?
* pmichaud thinks it would be an excellent idea
<chip> OK then ... I'll just look over the diffs by eye a bit for "oh no don't do it that way" stuff, which I doubt I'll find, and Allison gets a veto
<audreyt> woggle: very quick question - what's the state of "choice" "retry" "assert" in the STM branch? already in as primitives? planned? not in scope?
<woggle> audreyt: 'retry' support exists.
<woggle> audreyt: 'assert', too, I think, if I understand what you mean.
<chip> leo_: Next release is before OSCON if (in)humanly possible.  Say, Friday.  I might end up having to make it on Sunday
<woggle> audreyt: I _think_ you can do choice with the primitives I have.
<leo_> then merge can only be thereafter and isn't a problem
<Coke> might it be better to hold release until after hackathon?
<Coke> then you have more to include, neh?
<audreyt> woggle: cool! will look into it in the Pugs/PIR backend after the release+merge. thanks a lot
<chip> ... and that would put us back on the monthly schedule
<audreyt> except the hackathon has a way of break trees in the past... :)
<chip> I'd hope hackathoners would be working from the trunk whether it's released or not
<chip> Anybody not doing that?
<leo_> woggle: [re clone] while at STM, you might try: if (PObj_is_PMC_EXT_FLAG(pmc)) thaw_freeze(..) else VTABLE_clone(..) 
<chip> <crickets/>
<chip> OTOH, I should finish pdd21 so we don't release with it half-implemented.
<woggle> leo_: Ah, that sounds like something to try.
<chip> So I think that puts it after OSCON.  <<--- contradicting previous stmt
<audreyt> does that means we get STM partially merged before 0.4.6? :)
* pmichaud always works from the trunk
<chip> audreyt: yes
<chip> (probably)
<obra> ok. anything else?
<leo_> woggle: would that early merge impact the outcome of the Soc project?
<woggle> leo_: So long as I can still work in a branch (when writing new, fabulously broken code), probably not.
<leo_> great, sure, he branch can continue
<obra> any last questions?
<obra> failing that, schedule: we _Will not_ meet during OSCON
<obra> see y'all in a couple weeks
<audreyt> obra++ # thanks
<leo_> obra, moderator - thanks guys - see ya
<chip> ciao
* chip adjourns to #parrot
% woggle has left #parrotsketch
% pmichaud has left #parrotsketch
% leo_ has left #parrotsketch
% Coke has left #parrotsketch
% particle has left particle!~particle@144.81.84.193
% particle has joined #parrotsketch
% particle is now known as ^particle^
% ^particle^ is now known as particle
% particle is now known as [particle]
% [particle] is now known as particle
% Psyche has left Psyche!~Psyche@e176067244.adsl.alicedsl.de
% Psyche^ has joined #parrotsketch
% Psyche^ is now known as Psyche
% allison has left allison!~chatzilla@sub17-30.member.dsl-only.net
% Psyche^ has joined #parrotsketch
% Psyche has left Psyche!~Psyche@e176067244.adsl.alicedsl.de
% Psyche^ is now known as Psyche
% vsoni has left vsoni!~vsoni@24-197-218-168.dhcp.roch.mn.charter.com
% Psyche has left Psyche!~Psyche@e176083084.adsl.alicedsl.de
% Psyche^ has joined #parrotsketch
% Psyche^ is now known as Psyche
% Psyche has left Psyche!~Psyche@e176083084.adsl.alicedsl.de
% Psyche^ has joined #parrotsketch
% Psyche^ is now known as Psyche
% Psyche^ has joined #parrotsketch
% Psyche has left Psyche!~Psyche@e176083084.adsl.alicedsl.de
% Psyche^ is now known as Psyche