see y'all later today % Coke has joined #parrotsketch % leo has joined #parrotsketch Hi, all Helo obra Hello. Hi Guys. We'll give everybody a few more minutes to appear. % chromatic has joined #parrotsketch hey chromatic Good day all. ok. we're 10 minutes in let's get into it leo: want to start? okie - div by zero exception incl. x86 and ppc JIT support - reworked OrderedHah.pmc based on insert stability of hash.c - named argument syntax proposal - experimental package-scoped closures - the usual bugfixes and maintainance work I hadn't too much time this week - that's all. ZZ ok. what are you currently blocking on? still chip and my time ok. I know he was around a bit last week. did you get some answers from him? not enough, no ok chip: are you here? % audreyt has joined #parrotsketch greetings. hey audrey chromatic: what's up with you and perl6? I wrote an article for Perl.com and published it last week. No code; still finishing Perl Hacks. it's a very nice article :) Thank you. ok. are you blocking on anything? Nothing three weeks of real life won't fix. ok audreyt: how's pugsworld? release engineering 6.2.11 was supposed to get bug triaging done today but sidetracked into various things one of them being implementing :closure for Parrot (as posted on p6i and my journal by leo's request -- some input from chip would be lovely) found out that IMCC already used up all private PMC bits for subroutines which complicates things PIL1->Parrot targetting is now fully regained we are back to Vienna hackathon status -- actually much better plan to work on PILN->Parrot following Lua and Tcl's model (porting our representation types to PMC) after 6.2.11 would also help if Parrot's Syck binding becomes two-way, as one of the things blocking &eval working on the parrot backend was there were no ways for Haskell/P6 and Parrot/P6 to share AST and other structures easily. not blocking on anything except time eor. ok. having fun? tremendously. Since I don't think chip is here.. we've got a bunch of time :) having a nonsegfaulting parrot is really really a big morale boost. Coke, want to give us a rundown about what you've been up to? partcl status: All tests pass again (OS X) - fixed the [expr] bugs that were causing some failures. And, Win32 is at 99.7%. This much improved status thanks to Jonathan. Converted [time] and [while] into templated builtins. Showed the [time] template to #tcl (freenode), and spotted that cvs-head behaved slightly differently and offered a one-line patch. jisom patched partcl to use the new Getopt::Obj. Tried to use RT for partcl in the same fashion I suggested it be used for parrot a while back: https://rt.perl.org/rt3/Ticket/Display.html?id=38113 is a milestone ticket for partcl. Biggest parrot-related blocker at the moment is namespaces (https://rt.perl.org/rt3/Ticket/Display.html?id=38118) We are now passing 5.74% (931/16210) of the converted test suite. ^Z other smaller blockers? time. There's plenty of small work to do atm, but the small stuff isn't as fun. =-) understood ok. I think we're up for questions unless I missed any regular participants who are here. chromatic: go for it Audrey, what do you mean by "two-way" bindings for Syck? chromatic: runtime/parrot/library/YAML/Emitter/Syck.pir Looking now. also syck_parse doesn't roundtrip objects. but even a simple emitter would help a lot. Is this file in the Parrot or Pugs repo? that file is in the time machine's scope sorry, I didn't make myself clear. I was referring to a hypothetical thing that I'd like to have. Just NCI bindings to the Syck lib? yeah a nice start would just to port what parrot's Data::Dumper could do but what I'd really like is to dump/load the ParrotObject structure which would make the PGE/TGE/whatever_else transparent to other languages. That might be more difficult. indeed... it amounts to sv_dump in p5 That probably needs C support in PMCs for introspection. which would be nice, too but a simple start that only does what Data::Dumper does would be quite nice already... I can retire PGE::Hs in the parrot tree then and use the far less fragile (and cyclic-happy) YAML for Pugs/PGE integration Makes sense to me. cool :) Is there a TODO filer in the room? chromatic: what do I mail where? % particle_ has joined #parrotsketch Coke: are you sleeping ;) TODO: add PIR bindings for libsyck No, just avoiding volunteering for work. Sure, I'll file it. hey particle_. coke gave a nice report about the state of tcl. anything you need to add? sorry i'm late. er. wait. I'm mixing people. sorry particle_: what's new and exciting? added test files for every pmc type now. just one test (.new) but it's a start to improve coverage that's all for last week. Coke++ # thanks. also TODO: PMC introspection API from PASM level (if it's not already there) TODO: create YAML::Emitter::Syck that does with Data::Dumper does. s/with/what/ chromatic: there is already PIR binding for libsyck's parsing part, fwiw How about for emitting? there is currently none There's the TODO then! :) ok. who else has questions? audreyt: how many Sub PMC flag bits would you need? And what for? leo: so POSTCOMP is already private7 and there is no private8 and I was intending to use it for :closure which is compile time autoclosing and I don't see a reusable bit nearby bit 1 is free isn't the :outer(..) enough to denote that it's a closure? so you mean all :outer should autoclose? I'm actually fine with that yes that's how it currently works huh you mean you implemented it today? r11212 in that case, how do you deal with reentrance? see http://www.nntp.perl.org/group/perl.perl6.language/24483 scroll down to "20 for sure" I had a nice (if brief) chat with chip last week. He'd been absent for family reasons but says things should be better now. And now I have his cell # to hassle him more effectively. I'll send out mail this weekend hassling folks for next week ;) anything else before we close out for the week? Audrey, how far is the "soon" PIL -> Parrot targeting going to go? What Perl 6 features can I use? chromatic: you can currently use the Vienna hackathon features, which is everything at 6.2.0 level, which is sans objects. in other words, perl4 audreyt: I presume that '&g' would be created with newclosure, as it needs :outer(f) chromatic: once PILN is ported, the object part should Just Work. chromatic: as for how soon, I'm aiming for Israel hackathon. which is one month from now leo: but you said the :outer(f) alone makes it autoclosing. 6.28 level support? chromatic: yeah, as it's the release date for 6.28.0 Lovely. audreyt: only along the call chaing, which is not the case here chain even I'm lost I thought autoclose is a compile time thing oh wait you mean if my outer happens to be my caller, it closes otherwise it does not? yep that sounds very bogus. { my $x; sub f { ... }; }; f(); f's outer is not its caller so I'd still need a newclosure? (and where, btw?) such cases would currently need newclosure, yes yeah. what I proposed was, because I can statically decide that the surrountine Block is not reentrant &f can autoclose to $x because there is no obvious place to do the newclosure-storage ok - I see I wonder if @ANON alone is sufficient I mean :anon (TODOs opened) Coke++ if my :outer is :anon, then autoclose, otherwise don't bother the :outer() is a better hint I think I'll fix the autoclose excellent leo++ I too would like :outer to autoclose always, so even in the first case in my p6l post, it'd work sub f ($x) { sub g { $x + 1 } } g(); would just +1 on an uninitialized register I'd take the consequence of whatever happened with that :) but the {my $x; sub f{}} case is more frequent, and more tricky to codegen, so fixing that would be nice already (eor) thanks everybody % chromatic has left #parrotsketch thanks obra % particle_ is now known as particle % particle has left #parrotsketch audreyt: sub f ($x) { sub g { $x //= 0; $x + 1 } } # else you'd get a Null access exception leo: that is exactly what I'd like. I prefer a Null access to a "cannot find $x" ok then woot leo: you going to GPW? yes mind if I stay in your place for a few days (say 4~5) afterwards? :) np, another hackathon ;) woot I'll book my return flight a week later then great % Coke has left Coke!~wcoleda@host-69-95-19-5.roc.choiceone.net % Coke has joined #parrotsketch % Coke has left Coke!~wcoleda@cpe-24-194-223-228.nycap.res.rr.com % Coke has joined #parrotsketch % Coke has left #parrotsketch