00:25 japhb joined 05:05 contingencyplan joined 05:47 contingencyplan joined 09:04 cognominal joined 09:07 cognominal joined 16:41 pmichaud joined 16:43 Auzon joined 17:01 japhb joined 17:22 Whiteknight joined 17:53 NotFound joined 18:06 tco joined 18:07 cotto_work joined 18:15 allison joined, barney joined 18:16 DietCoke joined 18:26 Whiteknight joined 18:28 chromatic joined 18:29 jonathan joined
DietCoke a big pre-start time hello to everyone. 18:30
moritz waves
Tene HICOKE!!!
DietCoke (barely!) Can I get a "hey!" from everyone who is conscious? =-)
18:30 jhorwitz joined
Whiteknight hey 18:30
NotFound Hey
allison hi
jhorwitz yo
jonathan hi
DietCoke wow. Big crowd today.
chromatic hello 18:31
DietCoke Ok. I reserve a question, but am going to report first this week. =-)
NotFound (for some value of 'conscious')
barney hey
18:31 wknight8111 joined
DietCoke * Switch perlcritic.t over to Test::Perl::Critic and use a perlcritic.conf file. Avoid reinventing the wheel. 18:31
* Fix bug regarding using 'cmp' and subclasses of 'Float'. opened a new ticket to track the more general case here.
* Add a new deprecated warnings flag.
* Add the ability to mark ops as deprecated.
* Avoid passing along -w to parrot from Test::Harness. (predates recent update by mpeters applied by chromatic)
* Delete very old, broken, obsolete pod_errors.pl
* General RT queue upkeep
I'm going to throw barney under the bus next while I figure out who we have on deck. =-) 18:32
barney is diving
Fiddled with Plumhead(Pipp).
Make most lib tests from fperrard++ pass with Parrot PCT 18:33
pmichaud hey.
barney Finished $DAYJOB project. Will have some Parrot time.
.eor 18:34
DietCoke excellent. Glad to see the work on PHP.
pmichaud, how about you?
pmichaud ** Rakudo spectest_regression: 79 files, 1240 passing tests (+13, +391 from 06-24) 18:35
== Overall
: Wrote my 3rd progress report for the Mozilla Foundation/TPF grant. Only one report left to go!
: Wrote an article about the Perl 6 test suites.
== Parrot stuff
: Jonathan++ made heroic efforts on lexical handling last week
: Jonathan++ added an implementation of :lexid
: Jonathan++ even fixed the "method cannot do :outer" bug (RT#47794)
: Unfortunately, lexicals are still a blocking issue for Rakudo (RT#56184)
: I created a meta-ticket (RT#56512) to keep track of the ticket dependencies
== PCT stuff
: Removed the :method :outer workaround that depended on RT#47794
== NQP stuff
: Played with trying to get lexical scopes and immediate blocks working correctly, leading to various tickets mentioned above
== Rakudo stuff
: Added a workaround to check arguments to zero-parameter functions
:: This is related to RT#39844
: Moved more builtins into the Any class -- still more to go
: Had a long discussion with jonathan++ about :instanceof() and HLL subclasses of .Sub
: Added the listop contextualizers
: Eliminated the Arrayref cheat -- now we just have an infix:<=> method cheat
: The remaining assignment cheats will depend on fixing infix:<=> parsing (which in turn wants some PGE refactoring)
: But this will also fix list assignment 18:36
: Added Order type
: Reported a Parrot bug in Float subclasses that caused exclusive ranges to fail
EOR
DietCoke (that last bug you mentioned is the one I fixed, right?)
pmichaud (yes.)
DietCoke (woot)
pmichaud (so add a DietCoke++ to it.)
DietCoke Ok... lets see, jonathan got a lot of press in your report, does he have anything of his own to report? 18:37
jonathan * Had Rakudo day last week; didn't feel like the most productive of them
* Got :lexid implemented, seems to mostly work, chromatic++ and NotFound++ contributed some fixes
* Chased down another few bugs related to lexicals...
* ...and then pmichaud found at lesat one more that so far I think is unresolved. I tried one different approach, which kept the initial thing I fixed working, but didn't help with the new issue.
* Also got a patch upcoming on Sub/Method differentiation. Maybe. Not completely happy with it yet, and it looks like there's some currnetly un-debugged Parrot issue to solve first. 18:38
* Really should have written my report for last week's Rakudo day...will get onto that soon.
* This week am *very* tied up with $other_job, but after this week it gets much quieter (because the time-critical phase of it is over and I can delegate a bunch of it elsewhere)
* Going to move this week's Rakudo day to next week and have two days then; much rather that than a Rakudo day this week when I pretty much know I'm going to get distracted by a bunch of stuff
EOR
pmichaud (moving Rakudo day to next week works _very_ well for me.)
jonathan (awesome...also two Rakudo days a week during July should be the norm, once this week is over)
RLYEOR 18:39
DietCoke Glad to see that parrot was pretty much able to take advantage of your rakudo day. =-)
Lessee. Tene's enthusiastic greeting should be rewarded with a spot in the first half. Go ahead.
Tene * Started adding block handling to lolcode. 18:40
* I think I can completely eliminate the ugly runtime lookup stuff.
* Queue one question for pmichaud.
KTHXBYE
DietCoke ooh, more lolcode. That makes me think we should jump to jhorwitz.
jhorwitz what a nice transition 18:41
no progress on anything practical last week
so i optimized for fun, finishing a usable version of mod_lolcode
fun with cheezburger.lol was had by all, and should make interesting fodder for OSCON
EOR
DietCoke Ceiling cat is pleased with your progress. 18:42
jhorwitz ceiling cat knows too much
DietCoke Why don't we switch over to Whiteknight, and then follow up with his mentor.
Whiteknight = GSOC_PDD09 work:
* migrated PMC_EXT and other simple objects to be managed by the GC
* More bug fixing and debugging on GC
* Build process mostly completes
* Passes all but about 600-700 subtests as of this morning
* Worked with DietCoke on enabling "self" in :vtable functions (still buggy)
One question to enqueue 18:43
EOR
chromatic Reviewed Whiteknight's branch, fixed some bugs there.
Fixed other bugs (memory management, GC, etc).
Applied some patches.
I want to get our bug queue down to 600 for the next release.
pmichaud 600 sounds ambitious. :-) 18:44
chromatic I also want to buy you all a space between the tokens RT and #\\d{4,5}, but that's just me.
I'll be around with more time for the next week and a half.
EOR
DietCoke Alrightee. cotto_work? 18:45
cotto_work * worked with Barney (and fperrad to some extent) to pick a new name for Plumhead
* the new name will be Pipp ("Pipp is Parrot's PHP")
* bought pipp.org, should have it by Friday
* have also been refactoring and increasing test coverage for phparray PMC
* misc other general Parrot fixes
.eor
DietCoke I should point out that technically, cotto_work did not commit those parrot fixes! 18:46
cotto_work nope, cotto_home did
DietCoke mmm, evil twin.
moritz, you're up. 18:47
cotto_work I've been very careful about that.
moritz Work on Rakudo:
* Added about 13 files to spectest_regression
* Added a tool for plotting test numbers
* One more tool in the making, this one to automatically unfudge tests
EOR.
DietCoke whoosh!
Alrightee. NotFound waved earlier, so I'll let him go if he wants to report anything. =-)
NotFound Not much, some patches, attempting to clean parrot debugger. 18:48
EOR 18:49
DietCoke That leaves our fearless Architect Chair. (from the 2nd edition monsters manual, I believe.)
chromatic I hope she uses her daily power of Answering Many Design Questions.
DietCoke 2 questions per level!
allison, go ahead. =-)
allison need to get chatzilla to flag "architect" 18:50
- Between travel and work, my Parrot time has been limited this week. (The next couple of weeks will be spotty, leading up to OSCON. After that I'll have lots of Parrot time.)
- I changed the "peanut butter" and "bad beer" opcodes to 'local_branch' and 'local_return', and added range checking to 'local_return'.
- Deleted the 'branchcc' opcodes, which won't be used.
- Fixed a problem with the code to delete an exception handler after it's invoked once.
- Between my fixes and chromatic's fixes, we're down to 4 test files with failures on the pdd25cx branch. 18:51
EOR
DietCoke Excellent. I am reminded to re-test tcl there. (say that out fast.)
chromatic I made Larry Page say that once.
DietCoke Ok. Anyone else have a report who didn't speak up?
chromatic++ 18:52
FYI, we are currently at 656 new/open tickets in our RT queue. I am sure we could eliminate many old ones that are just "what the heck is this comment?" See the parrotblog post I made a few weeks ago if you want to help close tickets but aren't a "parrot hacker". 18:53
My question is... Does IRC work as a meeting medium for all y'all?
(I will assume "yes" unless you squawk. =-)
Whiteknight it works for me 18:54
DietCoke This is kind of a self selecting question, of course. I know of at least one person who'd like to attend but can't wrangle irc during business hours.
So, I'll go with "works for most of us for now, anyway."
pmichaud it works very well for me.
(again, self-selection.) 18:55
DietCoke Tene: you had a question. Shoot.
NotFound This is not business hours for me.
Tene Actually, question retracted.
allison DietCoke: could they wrangle anything other than IRC? generally if a non-work IRC meeting is a problem, any other format would be as well
DietCoke well, that was easy. Allison, the other option would be phone, I think, ala sixperl. 18:56
allison it works well for me, as I often have internet access where I don't have phone access, but any textual meeting form would work for that
DietCoke Of course, then we'd lose our automagic-recording secretary and have to nominate chromatic.
pmichaud phone is less useful for me.
chromatic FAIL
moritz think that many non-native english speakers might have problems with phone. I certainly would
DietCoke fair enough. our sample size is large enough we can stick with IRC. works for me.
NotFound Agreed.
allison also, phone calls aren't free 18:57
DietCoke I have lived in world where they are free for so long I tend to forget that. =-)
So, IRC it is.
NotFound Better say, native english's ears will have the problem with me X-)
cotto_work irc++
DietCoke That leaves Whiteknight's question.
Whiteknight okay, I would like to add a new PObj flag to mark stack chunks 18:58
it looks like we've got space for one more in pobj.h, and I want to steal it for this purpose
allison we're going to have to add another set of flags soon, or remove some 18:59
this should be fine
Whiteknight Conversely, I could find some otherwise nonsensical combination of flags to use
allison though, why are you marking stack chunks? 19:00
Whiteknight Stack chunks are already being marked, see src/stacks.c:mark_stack.c
allison clearer question: why do you need a new flag for it?
Whiteknight I want to separate it out, only mark the head of the stack for the root set, and mark the remainder in the GC
it will give us a more clean iterative behavior 19:01
The GC i'm writing does tree-at-a-time marking, taking items from the root set 19:02
having fewer things in the root set means fewer trees, and fewer small trees, which means fewer iterations
allison would stack chunks be PObjs? 19:03
if so, why do they need different flags than other PObjs?
Whiteknight They are isomorphic with PObjs, and I'm treating all isomorphic objects like that together
allison if not, why do they need a flag in PObj?
isomorphic should mean they use the same GC flags 19:04
a tri-color mark needs 2 flags
Whiteknight right, i'm talking about other flags 19:05
chromatic Given two different structs (both isomorphic with PObj), after you've cast them to PObj, how do you tell what they originally were?
Whiteknight exactly, thank you chromatic
They are all getting marked with the same GC flags, but I need flags to determine what data type they are, and therefore where their children nodes are
allison you want a flag to identify that the original item was a stack chunk rather than a PMC?
Whiteknight yes, exactly
PObj_is_stack_chunk_FLAG or similar 19:06
allison fair enough, we're already littered with flags to tell if a PObj is a PMC, a class, an object, etc...
go for it
Whiteknight thanks, makes my life much easier (and hopefully will make a better GC) 19:07
allison we'll add another set of flags next time we need them
I still want to rip out the last stacks as soon as possible
Whiteknight You mean the dynamic_env stack? 19:08
allison yes
that's the last one remaining
Whiteknight oh, well if you are ripping that out eventually, there's no sense me making all sorts of special admissions for it
allison if it's a quick fix, make it
Whiteknight okay, will do. 19:09
EOQ
allison but, if it's going to take up a lot of time that could be spent on the main part of the gc, it's not worth it
Whiteknight it takes less time, I need to worry about fewer special cases
okay, that's all for me. DietCoke? 19:11
19:11 davidfetter joined
moritz seems we're loosing audience and ceiling cat ;) 19:13
particle well, then, i'll report status and we can all go home. 19:14
i had a chat with tewk yesterday about gsoc progress.
he's been using git instead of svn, but hasn't found a good way to update his branch consistently
chatted with mugwump about possible solutions, and will make that info available to tewk asap 19:15
turns out my desktop was brought down by faulty kb drivers which caused internal_power_error during windows boot
fixed that problem, now rebuilding my dev env 19:16
updating bundle-parrot with t::h 3
should have a cpan release today
.end
DietCoke Excellent. 19:17
I agree with particle's assessment that this is a good time to break. =-)
particle: once you have your dev environment back, do us a favor and go back through the old windows tickets. =-) 19:18
Anyone blocking on anything?
cotto_work legal, not that that can be helped
DietCoke (and for those reading at home, feel free to email the list or me if you're blocking on something. I may not be ceiling cat, but I'll see what I can do.) 19:19
pmichaud I'm blocking on lexicals, as mentioned previously.
DietCoke I hear there's going to be hacking at YAPC::EU; Good luck with that, wish I could be there, but it's not looking good. pmichaud: there was a proposed fix to the list, did you try it out? 19:20
moritz pmichaud: did you see chromatic++'s commit shortly before #ps?
pmichaud the fix for :lexid? Yes. That only fixes one of the issues.
RT#56184 is the real blocker. 19:21
DietCoke ok. 19:25
We sort of petered out there. =-) Everyone have a great week, patrick, let me know on #parrot if you stay blocked too long.
moritz ciao, everyone++ ;-) 19:26
allison thanks dc!
19:26 chromatic left 19:27 cotto_work left, NotFound left 19:28 allison left 19:34 Auzon left 19:38 jonathan left 20:15 particle1 joined 20:19 barney left 20:46 contingencyplan joined 21:20 jhorwitz left

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!