% [particle] has joined #parrotsketch
% [particle] is now known as particle
% jisom has joined #parrotsketch
% pmichaud has joined #parrotsketch
% chromatic has joined #parrotsketch
% bernhard has joined #parrotsketch
% allison has joined #parrotsketch
<particle> howdy all. coke's unavailable again today.
<chromatic> I'm happy to run things, or particle can.
<particle> go ahead, c.
<chromatic> allison?
<allison> - I checked in a first draft version of the PMC PDD.
<allison> - Spent the weekend working on ticket resolution and patch application.
<allison> EOR
<chromatic> bernhard?
<bernhard> Trying to get languages/lisp working again
<bernhard> .eor
<chromatic> I fixed a couple of bugs and applied a few patches.
<chromatic> I also cleaned up some code.
<chromatic> Then I ran out of time and went on a trip and realized that the release is today.
<chromatic> EOR
<chromatic> leo?
<chromatic> particle?
<particle> ~ i committed my queue of pending patches, since i was able to get all my dependencies fixed. this included namespace, exporter, object, and other work.
<particle> ~ i've turned my attention to Super PMC, to try to get it working. if anybody knows precisely what's broken, i'd appreciate comments.
<particle> ~ did some warnings cleanup today for release prep
<particle> cia is down, so i don't have more specifics.
<particle> .end
<chromatic> pmichaud?
<pmichaud> * I'm continuing to look at a PIR-less form of transformation for perl6 and other languages
<pmichaud> * Created a draft of the abc grammar using embedded actions, now available as languages/abc/src/abc-2.pg
<pmichaud> * Comments are welcome
<pmichaud> * Based on comments from #parrot, it's likely that we will separate actions from the grammar, and then PGE will call the actions as embedded closures at runtime
<pmichaud> * This will means that the actions can be written in any language, including the "light" version of Perl 6
<pmichaud> * So far the "light" version is very light -- thus far we only need scalars, arrays, hashes, $/, $<...> variants, method call syntax, and loops
<pmichaud> * Today I'm looking at contextual variables (needed for lvalue context) and how to handle the operator-precedence parser actions.
<pmichaud> * I expect to have running code by this time next week.
<pmichaud> EOR
<chromatic> tewk?
<tewk> Added t/pdd15oo/ tests that contain a copy of all test that use the newclass.
<tewk> the newclass opcode is replaced with newpdd15class opcode in those tests.
<tewk> The new pdd15 object.pmc doesn't call vtable entries that are overloaded in PIR.
<tewk> This functionality is usually implemented by installing delegate redirects into the vtable that
<tewk> find the PIR overloads and calls them.  Installing theses delegates should be done by Pmc2c.
<tewk> Pmc2c has organically evolved into a mess over time. So....
<tewk> I've spent the last week cleaning out Pmc2c. I have a detailed understanding of what Pmc2c
<tewk> tries to do. I've gone over every line. I'm getting close to being done.
* chromatic boggles
% mj41 has joined #parrotsketch
<chromatic> That's it for reports.
<chromatic> Are there active questions?
<particle> any dependencies for release today?
<chromatic> I note that a full test run on my machine takes under 5 minutes now, which is the fastest I've ever seen.
<particle> lucky you
<pmichaud> why faster?
<chromatic> The only dependency to my knowledge is the "Don't run codingstd tests in a release tarball."
<particle> hrmm
<chromatic> There was a vtable lookup fix a while ago that replaced a quadratic lookup with something substantially faster.
<pmichaud> hmmm, on my smokes it does appear that we've gone from 9.67 mins to 8.00 mins for "make smoke"
<chromatic> Not a huge win, but 10% isn't bad.
<pmichaud> 9.67min as of last Wednesday, 8.00 mins as of this morning
<chromatic> A few more 10% fixes would add up.
<pmichaud> of course, that's for a full make smoke, which tests a lot of things not really related to Parrot efficiency :-)
<particle> get me io, i'll get you a fast test suite
% davidfetter has joined #parrotsketch
<pmichaud> otoh, it appears that my system is now only passing 96.23% of tests, which is the lowest I've seen
<chromatic> I ran a fulltest last night and there were a few JIT failures that I have no idea how to fix.
<jisom> i386?
<pmichaud> x86_64-linux_ccache for those who are wanting to look
<chromatic> I disabled one test for the JIT, as there's no way it would pass.
<pmichaud> and my system doesn't run JIT
<jisom> pmichaud, do you get t/pmc/complex.t failures wrt 0.0 and -0.0?
<chromatic> Hm, those are all sprintf failures.
<chromatic> Perhaps we should move the debugging to #parrot.
<pmichaud> jisom: no, complex.t seems to pass all tests on my sys
<chromatic> Are there any other questions?
<pmichaud> I have no other questions
<pmichaud> I'm just studying and writing code today :-)
<particle> c: last week you said super has been broken a long time. do you know why?
<chromatic> I'm not sure the super opcode has ever worked.  I didn't realize we had a Super PMC.
<particle> or how? there's been no tests for it until i wrote some this week
<particle> there's a super opcode?
<pmichaud> afair, there's been no official support for "super" in Parrot
<pmichaud> there have been some workarounds
<chromatic> I think pmichaud and I have both worked around it.
<pmichaud> I've just not used it :-)
<particle> if anyone can help me work around it, i'd appreciate it.
<particle> i'm working on a pir subclass of Exporter PMC
<particle> method override, currently using Super PMC, is busted.
<pmichaud> is it the method override that is busted, or the need to invoke a method of the superclass?
<chromatic> My guess is method redispatch.
<pmichaud> right -- I never got method redispatch to work
<particle> when i redispatch, the object is undefined
<pmichaud> so I figured out how to do things w/o redispatch
<particle> somehow self is... missing
<chromatic> Did Alek's invoke() patch fix that?
<particle> iunno
<particle> it was borked on win32
<tewk> Aleks patch didn't get applied did it?
<particle> applied/reverted
<chromatic> It did, but it was whack-a-platform with regard to Win32/Unix for a while, so Allison yanked it last night before the release.
<tewk> I need to go look at his patch.
<particle> the issue was whitespace in a .pmc file, which is a bit maddening
<particle> s/opcode_t* /opcode_t */
<particle> before, it works on *nix only; after it works on win32 only
<chromatic> I hope that the updated Pmc2c will generate standards-compiliant code.
<particle> i hope it will accept standards-compliant code, too, so we can modify the .pmc files
<particle> i'm happy to work on that, too
<tewk> If it doesn't, it will be straight forward to fix.
<allison> I'm also not entirely happy with forcing the invocant into the call
<tewk> particle: I'll check into a pmc2c into a branch 
<particle> tewk: good news
<tewk> How do I create a branch svn cp trunk branches/pmc2c ?
<chromatic> kid51 has some tools that might help.
<tewk> link?
<particle> oh, all of languages/perl6's tests fail on win32
<pmichaud> that's odd
<chromatic> Hm, I can't find them.
<particle> actually, the harness reports them passing
<particle> but they all exit with non-zero code
<particle> -G to the rescue, of course
<chromatic> That's probably the Don't Eat Segfaults patch.
<particle> some crazy string-related gc problem, i suspect. haven't been able to track it down yet.
<particle> there's just too much code for me to step through and keep my sanity
% PerlJam has joined #parrotsketch
<chromatic> I suggest src/debug.c as our FOTW.
<chromatic> Any objections?
<particle> none here
<chromatic> Any other questions?
<chromatic> Or comments?
<pmichaud> none here
<chromatic> Pithy complaints?
<pmichaud> I'll be out of town May 17-20, but will continue to work on parrot while I'm gone
<chromatic> Thanks for participating, everyone.  Let's adjourn to #parrot.
<pmichaud> see you next week
% pmichaud has left #parrotsketch
% davidfetter has left #parrotsketch
% allison has left allison!~chatzilla@sub17-30.member.dsl-only.net
% chromatic has left chromatic!~chromatic@sub17-30.member.dsl-only.net
% chromatic has joined #parrotsketch
% chromatic has left #parrotsketch
% allison has joined #parrotsketch
% allison has left #parrotsketch
% [particle] has joined #parrotsketch
% particle has left particle!~particle@144.81.84.167
% jisom has left jisom!~jisom@74-134-230-123.dhcp.insightbb.com
% bernhard has left bernhard!~bernhard@p549A28EB.dip0.t-ipconnect.de
% [particle] has left #parrotsketch