% leo has left leo!lt@feather.perl6.nl
% leo has joined #parrotsketch
% jonathan_ has joined #parrotsketch
% jonathan_ has left #parrotsketch
% allison has joined #parrotsketch
% chromatic has joined #parrotsketch
% Andy has joined #parrotsketch
% barney has joined #parrotsketch
<chromatic> Okay, who's here today?
<Andy> I'm here, but I have my weekly staff meeting now
<Andy> oh wait, no!
<allison> here
<Andy> It's in half an hour
<Andy> whee!
<barney> here
<chromatic> Alright, let's start.
<chromatic> Allison?
* particle is here, with jonathan and coke's reports
<allison> do jonathan and coke first and I'll type mine
<allison> particle?
<particle> okie
<particle> jonathan:
<particle> Sorry, can't make the meeting. Spent today hacking Parrot, with the aim of getting PDD15 classes inheriting from PMCs working somewhat.
<particle> * Fixed clone method of NCI PMC - it lost a flag
<particle> * The Bound_NCI PMC was sufficiently out of date that it assumed the old calling conventions. Brought it up to speed.
<particle> * Fixed a bug in instantiating inherited PMCs.
<particle> * Wrote my thoughts on PMC/PDD15 stuff into PDD17 marked as a proposal
<particle> * Implemented and wrote a test for overriding of non-vtable methods and calling non-vtable methods inherited from PMCs.
<particle> * Then did the same but for the vtable methods this time
<particle> * Most of the way there with making it so DYNSELF.meth(...) in a PMC that has been inherited from by a high level class will call overridden methods.
<particle> * Aware I've introduced some warnings in Object.pmc, will fix later tonight when I'm back or tomorrow
<particle> On Thursday I'm leaving for some vacation, so won't be at PS or on the mailing list next week.
<particle> coke:
<particle> Hackathon went well. Got some volunteers late thursday, put them to work on checking our test coverage on PDD15.
<particle> Started re-org'ing and cleaning up the help.
<particle> Migrated a lot of the design-related tickets to launchpad.
<particle> https://launchpad.net/parrot - we're using blueprints and timeline for now. bugs and code aren't moving.
<particle> This will eventually drive a real RoadMap document.
<particle> Did some minor cleanup on tcl, a guilty pleasure.
<particle> have my eye on some perl6 work after jonathan's recent patches.
<particle> Reminder to all core people: if you see something that is slightly more than trivial to do, open a ticket. Concentrate on the bigger stuff, try to encourage more developers to contribute patches.
<particle> Also: check out the command line RT if you haven't already.  (see the wiki for details).
% Coke has joined #parrotsketch
<particle> here's mine, while allison types:
<particle> ~ committed a fix for a particularly hard-to-find gc bug exposed only in perl6 on windows. thanks to chromatic, jonathan, and especially Ron Blaschke for the effort in finding and fixing this one.
<particle> ~ added the ability to regenerate the main Makefile without rerunning all of Configure.pl -- 'make Makefile'. this expands on my previous work allowing the same for languages/ makefiles (see languages/perl6 for details.)
<particle> ~ found a new gc bug exposed only in perl6 on windows. currently tracking it down, and could use some help.
<particle> .end
<allison> The biggest thing this week was YAPC. A lot of positive response on the whole group of Parrot talks, and new interested contributor.
<allison> contributors
<allison> The hackathon was also productive. I managed to tear through a few roadblocks for Patrick, which is very satisfying. Also, George Wilson put together a fantastic graphic representation of the new object model, which makes it quite clear.
<particle> i'd like to see that.
<allison> I'm not sure whether to check it in to the parrot repository or website (question for later)
<allison> I still have several patches from the hackathon I'll clean up and commit today.
<allison> EOR
<chromatic> barney?
<barney> Still trying to grasp languages/lisp.
<barney> Trying to uncomment old bootstrap code.
<barney> .eor
* Andy scribbles on the whiteboard: Will there be a pre-OSCON hackathon?  Can there be?
<allison> Saturday after OSCON at chromatic's house
<chromatic> My report.
<chromatic> Fixed a memory leak in IMCC.
<chromatic> Started to port Pheme to HLLCompiler and PAST-pm.
<chromatic> This required a fix from Patrick to support namespaces.
<chromatic> I'm close to having it done, and it removes a lot of code.
<Andy> Grrr, I'm leaving Friday.
<chromatic> I think some of the string problems we have are GC problems.
<chromatic> EOR
<Andy> CAN I MAKE A REPORT TOO?
<chromatic> leo?
<chromatic> Okay, no Leo.
<chromatic> Andy?
<Andy> headerizer now handles statics in source files.
<Andy> Working on nailing down still more restrictions from GCC
<Andy> Working on getting a Solaris box up so we can use Solaris lint
<Andy> for the 20% that it does that GCC doesn't.
<Andy> (such as finding set-but-not-used vars and other analysis-heavy stuff like that)
<Andy> Starting docs on headerizer so people get the zen of it.
<Andy> Basaically, my life is finding seat belts to install.
<Andy> EOR
<chromatic> Alright.  Questions?
<Andy> Oh one more: I'm digging into valgrind errors
<Andy> valgrind finds lots of things to complain about, and I want to ensure that they're indeed OK, or fix 'em.
<Andy> I'm wary of the supression file that we have.
<particle> andy: headerizer docs are high on my 'want' list. parrot contributors should have a place to go to understand how to work with parrot source, and how headerizer affects coding styles/methods.
<Andy> particle: Understood.
<allison> documentation graphics files: do we want them in the Parrot repository?
<particle> allison: who can modify them?
<chromatic> In theory, Inkscape or any SVG editor can modify them.
<Andy> Oh, one other big thing: parrot and perl6 workspaces moved to perlfoundation.org, off of creaky old Rakudo.org.
<Coke> (document graphic files) wiki is fine short term.
<particle> i say put the svg files in the repo, and the image on the website.
<Andy> HJere's hoping that faster wiki = more wikiing.
<allison> particle: they were created with graffle, which is proprietary
<allison> lemme see if george gave me svgs
<chromatic> George gave me SVGs as a test.
<Andy> It's a proprietary format, but does that mean we shouldn't bother keeping 'em in there?
<allison> yup, I've got the final version svgs
<Andy> oh, if we have SVGs, then never mind.
<particle> if we have svg's, we can modify them from there. we don't *need* the original.
<particle> right.
<chromatic> I have a valgrind question.
<chromatic> What are you seeing, Andy, and how are you fixing them?
<allison> docs/svg?
<Andy> I haven't fixed anything yet.  I've only been investigating.
<allison> docs/pdds/svg?
<particle> allison: docs/pdds/images?
<particle> docs/pdds/extras?
<allison> particle: images makes sense
<Andy> chromatic: Various uses of ununitizlied memory, and mis-reads from buffers.
<Andy> I dont' have any examples handy.
<Andy> Some of it we know is caused by optimization in GCC
<Andy> but not all of 'em.
<Andy> ok, here's one...
<allison> I'll only check in the one main image as PNG, the rest as svg
<Andy> http://paste.husk.org/8646
<Andy> Note that that's under -G, so no garbage collection
<Andy> that's also with no optimization.
<Andy> If we're going to have suppressions under valgrind, then I want them clearly documented.
<Andy> which we're lacking now in tools/dev/parrot.supp*
<Andy> rather, parrot*.supp
<chromatic> You're looking at the uninitialized reads then, not the leaks?
<Andy> No, not leaks at all
<Andy> But the GC stuff causes ALL SORTS OF HYSTERIA with invalid reads and such
<Andy> not just uninitizlied.
<Andy> OK I HAVE TO 
<Andy> OK I have to go in 3 mins
<chromatic> Any other questions?
<barney> Can we put the book on the wiki ? 
% Andy is now known as AndyAway
<allison> barney: if someone wants to take the time to convert it, sure
<allison> barney: and then convert it back again
<particle> allison: any policy for book updates? all commits welcome?
<allison> there are probably automated tools somewhere to do POD to ST's wiki format and wiki to POD conversions
<allison> particle: all commits welcome, it's part of the parrot distribution now
<particle> thanks.
<allison> (all commits very welcome, as it's quite out of date)
<particle> why is windows so good at finding gc bugs in parrot?
<particle> i wonder if it somehow does gc runs more frequently
<chromatic> Perhaps it uses more memory.
<particle> http://rt.perl.org/rt3/Ticket/Display.html?id=43507 is the new gc bug i referenced in my report.
<particle> one of the hardest things about gc debugging is that (like here) it doesn't show up in the backtrace.
<particle> is that *sometimes (like here)...
<Coke> if the svgs are *part* of the pdd, put them in the pdd directory. if they're just supplemental, put them in the docs directory for now. not /svg; we will probably move them anyway once I figure out how I want docs/ organized.
<Coke> (gc) is there anything we can do to make --gc-debug a more useful GC debug tool?
<particle> heck, i just wish i could tell what sub i'm in when debugging this code.
<allison> coke: I don't want to dump them straight in the docs directory, there's too many of them
<chromatic> I think all it really does is explicitly frees memory at program end so as not to give Valgrind lots of false positives.
<particle> a sub's name is buried so deep in it's structure, i can't find it.
<Coke> allison: ok, then use /svg for now.
<allison> only one is directly relevant to the PDD, the rest are examples. So maybe I'll just check in the one that's immediately relevant
<Coke> if you're just checking in one, doesn't matter where it goes. I compel you to check it in.
<barney> What are the plans for compilers/pct?
<particle> pmichaud holds that plan. it involves consolidation of current compiler tools. i don't have the specifics.
<allison> it's a fork and refactor
<barney> ok
<allison> and will replace the current compilers/past-pm
<particle> i'm having trouble increasing coverage on pdd15 tests
<particle> i don't know what's done yet. we don't have the tools to tell us.
<particle> i'm afraid we won't have them for some time.
<allison> what tools can we add?
<particle> so the only way to increase coverage is to try to implement something using pdd15.
<allison> or, what are the tools we can add now
<allison> we're not quite there yet
<allison> first we need to pass the test suite in t/pdd15oo
<particle> i'd like code coverage tools for pir.
<allison> then we can have patrick start porting pge, which should give use lots of test cases
<allison> s/use/us/
<particle> i'd like spec coverage tools (eg smartlinks)
<allison> particle: makes sense
<particle> patrick has already started porting pge. jonathan removed one blocker today (i'm pretty sure)
% smash has joined #parrotsketch
<particle> unfortunately, patrick's work isn't in the repo, and he's not around.
<particle> okay, i'll see what i can do about getting some t/pdd15oo tests passing.
<Coke> He'll be back... tomorrow, was it?
<Coke> particle: conversely, I'd like more tests written, too.
<Coke> (there's a ticket from the hackathon pointing out stuff that's missing.)
<allison> particle: there's a doc on the wiki about what tests still aren't passing and why
<particle> coke: i want more tests. i'm trying to unit test the pmc api.
<Coke> also, be nice to clean up those tests so they are JUST testing pdd15oo - there's a lot fo duplication.
<allison> http://www.perlfoundation.org/parrot/index.cgi?pdd_15_remaining_features
<particle> as i see it, the pdd15oo tests are functional tests.  we need both, surely.
<allison> the pdd15oo tests are temporarily forked tests
<allison> the real core tests are in t/oo
<allison> t/pdd15 will go away
<particle> when and how? do we get all of pdd15oo passing, then replace t/oo?
<allison> the pdd15oo tests are all copies of other tests throughout the system
<allison> (the original file name is noted at the top of the file)
<particle> ah, great.
<particle> i'm still not sure how we go about fixing things. do we modify the pdd15oo test files *only*, or the originals, too?
<particle> oh, allison, i need to talk to you about purify.
<allison> particle: ok
<allison> particle: we modify only the pdd15oo test files for now
<particle> chromatic: got a little time to help me with the latest segfault this week?
<chromatic> Should be able to tomorrow.
<particle> fab. i suspect i'll have time then too.
<particle> okay, i'll start attacking the 'tests no longer valid' from the wiki page. that ought to increase the signal/noise ratio.
<chromatic> It would be nice to have a release with substantially more feature tests.
<particle> i agree. there is no lack of opportunities when working on parrot.
<particle> my plan is this: study the pdd15oo tests one by one, and determine what features of pdd15 aren't adequately unit tested. beef up unit tests, and fix implementation where necessary.
<chromatic> TODO tests please!
<particle> yes, of course.
<particle> think we could code up a way to tell the test harness to make todo tests ignore that flag?
<chromatic> Who made the default harness report TODO tests, by the way?
* particle doesn't know
<chromatic> Good feature.
<particle> i didn't realize they were reported.
<chromatic> Since Thursday or Friday I believe.
<particle> crap-o.
<particle> i have tests that block when i use perl t/harness, but run fine with prove
<chromatic> t/harness sets some flags that prove doesn't, I believe.
<particle> when i attach the debugger, it says "the process appears to be deadlocked (or is not running any user-mode code). all threads have been stopped."
<particle> i'll check the os vendor site for bugs.
% AndyAway is now known as Andy
<Andy> wow, you guys are still at it.
<Andy> Did we want to talk about the zen of shims and interps?
<particle> things have wound down here. why don't you stick your thoughts in a doc or wiki when you get a chance and ask the list for review when it's ready.
<Andy> 'cuase it's more fun to just tromp through code with a broad axe.
<Andy> whack! slash!
% Andy has left #parrotsketch
% chromatic has left chromatic!~chromatic@sub17-30.member.dsl-only.net
% barney has left barney!~bernhard@p549A0F6D.dip0.t-ipconnect.de
% allison has left allison!~chatzilla@63.105.17.30
% particl1 has joined #parrotsketch
% particle has left particle!~particle@c-24-19-12-148.hsd1.mn.comcast.net