% particle has left particle!~particle@c-24-19-12-148.hsd1.wa.comcast.net
% particle has joined #parrotsketch
% Coke has joined #parrotsketch
% Coke has left #parrotsketch
% mdiep has joined #parrotsketch
% allison has joined #parrotsketch
% barney has joined #parrotsketch
% chromatic has joined #parrotsketch
% pmichaud has joined #parrotsketch
% jonathan_ has joined #parrotsketch
<chromatic> Hello everyone.  Who's here?
% mdiep has left mdiep!~matt@192.42.249.97
<pmichaud> here
<jonathan_> meee!
<barney> bernhard is here
<allison> here
<chromatic> Okay, let's get started.
<particle> hear hear
<chromatic> allison?
<allison> I've been working on the object implementation
<allison> I'm just about to check in a bunch of tests that were failing because of changes to the main line that weren't ported over to the pdd 15 test directory.
<allison> And, I've got a list of missing features and the failing tests that correspond to them, so I can distribute out implementation tasks.
<allison> I'll make an announcement today about the bug squashing session this Saturday.
<allison> EOR
<chromatic> barney?
<barney> Spent a long weekend in the Alps, without notebook.
<barney> Still playing with languages/lisp.
<barney> Changed IRC nick to 'barney'.
<barney> .eor
<chromatic> I fixed some bugs, particularly memory leaks and GC bugs.
<chromatic> I also poked at languages/perl6.
<chromatic> I'm going to profile it later and see if there are any easy optimizations.  Otherwise, I need to write documentation and I'll poke at whatever easy things can get the Perl 6 sanity tests passing.
<chromatic> jonathan_?
<jonathan_> * Checked in a document to docs/dev/ about PARROT_API to help people break the Windows build in one less way.
<jonathan_> * Started implementing PMCProxy PMC
<jonathan_> * You can now inspect any PMC like a PDD15 class: name, namespace, and its non-vtable methods. Other things still todo.
<jonathan_> * $P0 = get_class 'ResizablePMCArray' will now return a PMCProxy
<jonathan_> * Then $P1 = new 'Class' / add_parent $P1, $P0 to inherit from the PMC
<jonathan_> * If you do that, the PMCProxy sits in the parents list and MRO
<jonathan_> * On object instantiation, also instantiates and stores the any PMCs we inherit from, either directly or anywhere in the MRO
<jonathan_> * This should mean we can multiple inherit from PMCs, once this is done
<jonathan_> * Next is starting to make the inheritance actually mean something
<jonathan_> * Got a halfway there patch for delegating non-vtable methods
<jonathan_> .end
<chromatic> How about super?
<jonathan_> Still todo, but I'm not sure how that's done just for normal PDD15 objects
<jonathan_> Syntactically
<chromatic> particle?
<jonathan_> I probably know how to implement it, if I know what's wanted. :-)
<particle> ~ most of my work has been behind the scenes this week.
<particle> ~ trying to get parrot to compile on win32 x64. *my brane hurts*
<particle> ~ this has led me to refactoring configure.pl to support cross-compilation. that work is progressing, and should be committed soon
<particle> ~ ...that is, soon after i fix the gc bug preventing perl6 tests from running on win32. this is my top priority, i want it fixed before release. it seems an invalid pmc has been put on the end of the free_list (if i'm right.)
<particle> .end
<chromatic> pmichaud?
<pmichaud> * Congrats to Coke++ and chromatic++ for bringing perl6 down to only two failing tests in 01-sanity.
<pmichaud> * I've worked on Perl6Regex and Perl6Grammar a fair bit.
<pmichaud> * About to check in an updated abc.pg, that uses the new S05 quoting syntax, and makes token names more consistent with Perl 6 syntax (i.e., infix:<+> instead of 'infix:+' )
<pmichaud> * Should have that ci'd in about 10 minutes (running tests)
<pmichaud> * Next I'm going to see about updating the perl6 grammar to use the new syntax
<pmichaud> * Then I'll see if I can get the Bool::True test passing, and other issues dealing with namespace and object handling in perl6
<pmichaud> * Then back to working on the Perl 6 transformation subset, which is now substantially easier since Coke++ implemented a simple version of 'for' loops in PAST-pm
<pmichaud> eor
<chromatic> Did I miss anyone who wants to report?
<chromatic> Alright.  Who has questions?
<particle> where's leo when you need him? debugging gc is a great learning experience i wish i didn't have to do
<barney> Has Parrot something like heap inspection? Get a list of live PMCs, Strings ...
<chromatic> In theory you could run just the mark of the M&S.
<chromatic> In theory.
<pmichaud> (updated abc.pg just checked in, r18955)
<jonathan_> One thing that may help, is some way to get pmc2c to (optionally) emit a check a the top of every method in a PMC to make sure the PMC isn't on the three list, and blow up if it is.
<jonathan_> uh
<jonathan_> free list
* jonathan_ should learn English.
<chromatic> No, don't make that assumption.
<jonathan_> Of course, that may still not catch it.
<chromatic> The free list flag is only valid *during* GC.
<particle> jonathan: you just concentrate or your c, we'll translate to english
<jonathan_> chromatic: I think that's the case for the live flag; the free list flag is valid whenever something is on the free list, I thought?
<jonathan_> That was my reading of pobj.h, anyway.
<chromatic> Hm, I think you're right.
<jonathan_> Or wherever I read it.
<jonathan_> Is your understanding that whenever something new is created, we should always set the live flag, because we may be between a DOD and a GC run?
<chromatic> No... that would cause leaks.
<chromatic> I think you're right.
<chromatic> I was thinking of the free flag.
<jonathan_> OK.
<chromatic> I'm not sure checking non-free will get us anything, but if the assertion fails, it'll tell us something.
<jonathan_> I'm still trying to get all of this stuff into my head. :-)
<jonathan_> Well, the free list corruption shows up when a PMC that is on the free list gets modified.
<jonathan_> The first pointer is used for the PMC to hang its data off, but also where the linked list pointers are.
<chromatic> Something thinks that it's free but something else has a pointer to it and should have marked it live?
<jonathan_> Yes. DOD thinks its free and freed it, but it's really still being used.
<jonathan_> And in being used, it scribbles over the freelist pointer.
<jonathan_> And then get_new_free_object or whatever its called, which just pulls stuff off the free list, segfaults.
<particle> sure, cause it tries to pull something from 0xffffffff
<chromatic> Which is the int_val of an expired iterator in a Key.
<particle> indeed, it is
<chromatic> That makes me want to look at what goes into PMC_int_val(key).
<jonathan_> chromatic: I concluded the same and spent an hour with Key.pmc and turned up a blank.
<chromatic> Before we get too sidetracked, are there other questions we should answer?
<jonathan_> I may (hopefully) have missed something.
<chromatic> allison, do you have a list of tasks for bug day?
<allison> yes, they're in the ticket for 0.4.13
<allison> I'll put links in the mailing list message
<chromatic> Oh, here's a good question.
<chromatic> Is anyone blocking on anything besides time?
<particle> you're blocking on rt bugadmin
<chromatic> Yes.
<jonathan_> I'm blocking on understanding how to make inheritance from PMCs work sanely and well.
<jonathan_> allison: Any comments on the stuff I've put in this week, would be very helpful. I think it's along the lines of what you're after, but confirmation is always nice.
<allison> jonathan: write up a quick summary of as much as you've wrapped your head around so far
<allison> I'll look over the code, but the why is helpful
<allison> and, we can kick it back and forth
<jonathan_> Well, the code is mostly just PMCProxy, which is the bit I understand. :-)
<jonathan_> Anyway, will do.
<allison> many thanks :)
<chromatic> It sounds like we're wrapping up.  Is there anything else?
% mdiep has joined #parrotsketch
<particle> one question
<chromatic> Go ahead.
<particle> are the pmc methods that warn during pmc2c.pl fro default.pmc deprecated?
<particle> how do we get rid of the warnings?
<chromatic> I think we just delete them.
<jonathan_> If the warnings mean what I *think* they mean, it's dead code.
<chromatic> I whacked them once but didn't commit and all tests passed.
<jonathan_> Aye.
<particle> okay, then, somebody should do that someday.
<particle> is rt back up?
<allison> standard deprecation cycle
<allison> put the deprecation note in now, list it as happening for the 0.4.14 release
<particle> wow, no, it's not up.
<allison> particle: no, it's not
<particle> any timeframe? will it be up before bugday?
<barney> Is it backed up?
<allison> it is backed up
<allison> particle: they still say it'll be up today
<jonathan_> allison: If I understand correctly, the vtable methods were already deprecated and removed, and the code we'd be deleting hasn't been included in the generated .c files since then.
<particle> jonathan_: that's my suspicion, too
<chromatic> Agreed.
<allison> the nice thing about having a standard deprecation policy is, it's standard :)
<allison> do you want to delete them for the release on tuesday?
<jonathan_> It'd get rid of the warnings, and deleting it can't make any difference.
<allison> if so, do it now, so we have plenty of time before the release
<allison> don't do it Monday night
<chromatic> Monday night is for GC changes.
<allison> ggrrrrrr ;)
% allison has left allison!~chatzilla@c-24-21-59-142.hsd1.or.comcast.net
% barney has left barney!~bernhard@p549A26F1.dip0.t-ipconnect.de
<pmichaud> j/part
% pmichaud has left #parrotsketch
% mdiep has left mdiep!~matt@192.42.249.97
% tewk has joined #parrotsketch
% particl1 has joined #parrotsketch
% particle has left particle!~particle@c-24-19-12-148.hsd1.mn.comcast.net
% particl1 is now known as particle
% particl1 has joined #parrotsketch
% particle has left particle!~particle@c-24-19-12-148.hsd1.mn.comcast.net
% chromatic has left chromatic!~chromatic@sub17-30.member.dsl-only.net