Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GC tuning | remove deprecations |
Set by moderator on 21 October 2010.
dukeleto moritz: yeah, i noticed that. I always use noscript. 00:00
mikehh moritz: re ticket #1834 - looks like the old version is still there, when you create a new version, the install does not remove the old version there 00:01
moritz: if you look in the install dir you probably still have both 00:02
moritz mikehh: yes, but then why the error?
mikehh: it loaded one, so it must have found it. Why complain about another one?
dukeleto kid51: re: your question about the ipv6 config step. I think that branch can be merged 00:03
00:03 dngor_ left 00:05 dngor joined
kid51 dukeleto: So, does it do enough re determining ipv6 capabilities for you to do what you need to do? 00:06
00:06 theory left
kid51 dukeleto: Note, I just transformed kurahaupo's code into a config step -- but I have no expertise that would enable me to evaluate his code one way or the other. 00:07
mikehh moritz: yeah, I seem to remember getting some strange errors a little while ago where it was picking something up from an installed dir when I should have been working in the build dir
kid51 dukeleto: Actually, if you comment on those questions in the ticket, that would be best.
00:08 kid51 is now known as kid51_at_dinner
moritz mikehh: I don't really understand the issue at hand (and I removed the build dir, so I can't investigate more), I just say that the error message doesn't make sense to me, as is. Maybe a simple rephrasing would fix it, maybe the problem is deeper 00:09
second attempt at sleep&
00:10 dmalcolm left
mikehh moritz: I think it is a deeper problem, before I just used to remove libparrot.so to avoid conflicts 00:11
00:11 dngor_ joined 00:12 dngor left 00:29 dngor joined 00:31 dngor_ left 00:47 theory joined
kid51_at_dinner nick kid51 00:50
00:50 kid51_at_dinner is now known as kid51
dalek rrot: r49628 | jkeenan++ | trunk (2 files):
Per �trac.parrot.org/parrot/ticket/1785, complete a deprecation by changing a 'warn' to a 'die' in lib/Parrot/Pmc2c/PMC.pm.
 00:55
cotto ~~ 01:01
dukeleto cotto: howdy 01:04
cotto whiteknight, nice job blogging. 01:18
GeJ kid51: once llvm-gcc is installed, Configure.pl catches it correctly. 01:23
kid51 GeJ: thanks. 01:24
dalek rrot: r49629 | mikehh++ | branches/generational_gc/src/gc/gc_ms2.c:
[generational_gc] add missing c function docs
 01:26
kid51 GeJ: by the way, are you having any problems on freebsd with the issues I'm having on Darwin (TT #1829)?
01:27 dngor left
rrot: r49630 | mikehh++ | branches/generational_gc/src/gc/gc_ms.c:
[generational_gc] correct c function docs item
rrot: r49631 | jkeenan++ | failed to fetch changeset:
Merge tt1824_ipv6_configure branch into trunk. Add configure step auto::ipv6, associated probe and test.
01:29 dngor joined
GeJ kid51: will read. 01:29
dalek rrot: r49632 | jkeenan++ | branches/tt1824_ipv6_configure:
Branch has been merged into trunk and is no longer needed at HEAD.
 01:42
01:50 Hunger left 01:52 luben left, jnthn left, jnthn_ joined, he left 01:53 he joined
dukeleto kid51++ # merging the ipv6 config step branch 01:56
01:56 Hunger joined, davidfetter left 01:58 whiteknight left, luben joined
GeJ kid51: in darwin/sysmem.c could you please s/HW_MEMSIZE/HW_PHYSMEM/ and re-enable parrot/sysmem.h in src/gc/gc_ms2.c 02:04
?
kid51 Will try. 02:06
GeJ and see if this fixes your build?
cotto Both are valid according to Apple. I hope that's not what ends up being the fix, since it'd say bad things about Apple's docs. 02:14
GeJ cotto: agreed, but kid51's box is 6 years old. Plus, didn't Apple stop releasing new version of OS X on PPC after 10.5? 02:18
02:18 Andy joined
kid51 cotto, I hate to tell you this, but Apple's got a problem with docs. GeJ++'s suggestion worked! 02:24
GeJ ok, now we have another problem.
kid51 Of course, we'll need additional testing on Darwin/Intel, 10.5 and 10.6.
GeJ and 32 vs 64 bits
kid51 Yes. But I think we should commit this and encourage other smokers. We'll find out more quickly that way. 02:25
cotto I guess they spent all their tuits on ui and app store rejections 02:33
kid51 Good, maybe now I can start thinking about the Parrot project once again. 02:35
kid51 must sleep
02:35 janus left
cotto If it's defined, HW_MEMSIZE should be used. HW_PHYSMEM is limited to 32-bit values (though it'll be the max if memory is >2^32 B). 02:36
dalek rrot: r49633 | jkeenan++ | trunk (3 files):
Per suggestion from GeJ++, s/HW_MEMSIZE/HW_PHYSMEM/ to avoid build error on Darwin. This needs to be tested on Darwin/i386 in OS X 10.5 and 10.6.
 02:44
02:48 janus joined 02:51 kid51 left
dalek rrot: r49634 | mikehh++ | branches/generational_gc/src (2 files):
[generational_gc] some casts to get g++ to build
 03:15
05:38 tadzik left 05:39 theory left 05:49 Andy left 05:52 jhelwig left 05:55 chromatic left 06:06 fperrad joined 06:19 jhelwig joined
cotto I have a blog: reparrot.blogspot.com/ 06:20
06:21 mikehh_ joined 06:25 mikehh left
dukeleto cotto++ 06:39
cotto: what does "The tl;dr is that we'll be splitting responsibilities into 5 teams, only one of which will cover architecture." mean ? 06:40
cotto too long; didn't read 06:41
I'm fixing that now
dukeleto cotto: so is our team structure "officialish" ? 06:42
cotto "officialish" is a good word 06:43
I should clarify that.
dukeleto cotto: i volunteered for community management lead, but I am not so sure that it is "official". Am I the community manager lead? Do we need a vote or something? 06:45
cotto At this point, "official" is a pretty strong word for anything. 06:47
I suppose a vote would generate discussion and make it seem less like a handful of developers were forcing a structure on everyone else. 06:48
dukeleto cotto: sure. That is why I use "officialish" ;)
cotto (which isn't the intent, of course)
dukeleto cotto: yep, that is what I am hinting at. More legitamacy. I would like to see if parrot-users agree that our new structure is viable. 06:49
cotto I want to get things moving and part of me feels like a vote would just slow things down, but we're not working in a vacuum and the feedback we could get might actually get us further toward a stable structure. 06:50
dukeleto I was talking with some RTEMS developer tonight at the hotel and I will be talking with them more this weekend about how to make Parrot on RTEMS viable. They seem to be very excited about getting dynamic languages in real-time environments.
cotto: i am saying that we continue as if we are official, but do a vote as confirmation 06:51
cotto: the vote should not slow anything down
cotto I also don't want to do a vote if there's no possibility it'll change anything. 06:52
dukeleto cotto: i hear that.
cotto: shouldn't chromatic be on the architecture team on the wiki?
cotto We can say that we're the interim leads until a formal vote can be taken, at which time anyone may be put forward for a role.
I haven't asked him yet. 06:53
dukeleto cotto: i thought he said yes to that at the pdx gathering, but I could be wrong.
cotto My memory is treacherous, so it could have gone either way. 06:54
I don't remember it though.
dukeleto cotto: yep. I will ask him in person tomorrow :)
cotto wfm
Do you have the tuits to get the vote set up at some point within the next couple weeks? 06:55
dukeleto cotto: i think smash is the person to do that, if we wants to volunteer
cotto seen smash 06:56
aloha smash was last seen in #perl6 13 days 15 hours ago saying "colomon: i haven't had the time to improve the harness yet :(".
dukeleto cotto: he seems to be active on email, but not as much on irc
dukeleto goes in search of late-night food options 06:57
cotto happy hunting 06:58
bacek_at_work ~~
cotto hi bacek_at_work 06:59
bacek_at_work aloha, cotto 07:00
seen kid51
aloha kid51 was last seen in #parrot 4 hours 25 mins ago saying "Good, maybe now I can start thinking about the Parrot project once again.".
cotto particle, thanks for the b-day wishes 07:07
07:11 contingencyplan left 07:15 janus left 07:41 jsut joined 07:46 jsut_ left 07:57 wagle_ joined 08:02 wagle left 09:21 mikehh_ is now known as mikehh
mikehh opbots, names 09:21
09:23 jhelwig left 09:26 zostay left
dukeleto aloha, opbots? 09:26
aloha dukeleto: I give up.
09:26 zostay joined 10:45 dngor_ joined 10:46 oyse joined 10:47 dngor left 11:12 dngor joined 11:13 dngor_ left 11:17 dngor_ joined 11:18 dngor left 11:39 whiteknight joined
whiteknight good morning, #parrot 11:49
GeJ good night, whiteknight. 12:08
whiteknight hello GeJ, how are you doing? 12:09
brb 12:10
12:10 whiteknight left 12:11 whiteknight joined 12:36 bluescreen joined 12:42 Andy joined 12:49 contingencyplan joined 13:23 Andy left 13:56 fperrad left 14:01 brianwisti joined 14:24 jhelwig joined 14:47 mikehh left 14:58 tadzik joined 15:15 allison left 15:24 dmalcolm joined 15:26 mikehh joined 15:44 theory joined 15:55 dngor_ is now known as dngor 15:57 oyse left
dukeleto 'ello 16:11
brianwisti yo 16:13
dukeleto brianwisti: wazzup 16:15
brianwisti: what's next to do with parrot-handler ?
brianwisti Finish making Installer smart so I can make Handler stupid. 16:16
Specifically, I'm working on the build stages in Installer (`make test` and `make install` still waiting for me). After that I chop the heck out of Handler so that it relies on Installer to tell it what just happened. 16:17
16:27 cognominal left 16:34 cognominal joined 16:35 dip left 16:42 Andy joined 17:01 estrabd left 17:02 estrabd joined 17:09 davidfetter joined
cotto ~~ 18:08
brianwisti Confused at line 1, near "~~\\n" 18:09
cotto ENOTSMARTENOUGH
brianwisti Oy, tests take a long time when you're testing your app's ability to build and make *other* tests. 18:10
cotto Mmmm. Metatesting.
brianwisti It's metatasty. 18:11
dukeleto mmmmm, metagastronomy 18:14
moritz megaastronomy! 18:15
cotto throws some meta on the bbq 18:17
dukeleto metaburgers for everyone! 18:18
brianwisti I already had lunch. A nice parastrami sandwich on wry. 18:19
18:21 patspam joined
dukeleto what are people hacking on today? 18:44
moritz other people :-) 18:46
brianwisti $work and parrot-handler.
moritz actually most of my hacking time today was spent on my new vocabulary coach 18:47
(written in Perl 6)
(I'm trying to learn Norwegian these days .-)
jnthn_ plans to hack on exceptiony stuff in 6model 18:51
moritz like, exceptional stuff? :-) 19:01
jnthn_ Like, exceptions. :P 19:02
I did .leave yesterday.
dukeleto This is quite interesting: aws.amazon.com/free/ <-- free S3 and EC2 for a year 19:03
jnthn_ So can now implement return exception handling in terms of it. :-)
And yes, I'll do the handler search lexically, per spec. :-)
jnthn_ hopes that Parrot's exception model may come to look more like what he's in the proccess of implementing.
whiteknight jnthn_: Parrot's exception model does need an overhaul 19:11
if you would like to share notes...
jnthn_ whiteknight: *nod* 19:14
whiteknight: Will do - I probably need to write an implementation first to get more of a feel for it.
whiteknight: But if it works out nicely I'll write up the details. 19:15
wagle_ are the scripts in tools/dev/* documented anywhere (in addition to the stuff in each script file)?
jnthn_ whiteknight: The essentials of it are that stack unwinding isn't done unless/until needed, exception handlers are run in the dynamic scope of the place the exception was thrown, and thus resumable exceptions are simply provided by not causing a stack unwind. The whole model is free of needing to deal with continuations. 19:18
whiteknight that does sound nice 19:28
19:54 whiteknight left 20:32 lucian joined
Tene jnthn_: working on exceptions now? Even more tempting... 20:50
jnthn_ ;-) 20:51
Tene: Well, prototyping it in the .Net impl first.
It's an easy playground. :-) 20:52
Tene: If you want to look at enabling the same bunch of things on Parrot though, that'd be awesome.
Tene jnthn_: If you show me what you're interested in or need or want, I'd love to consider that. 20:53
Parrot's exceptions system is... suboptimal.
I don't actually have anything planned for tonight yet...
jnthn_ Tene: One thing that'd let us immediately fix up a long-standing Rakudo bug is a way to throw an exception lexically. 20:55
Tene walk the outers instead of the callers?
jnthn_ Right.
Tene Sure, that's totally doable.
jnthn_ One thing that I'm also doing (as per Perl 6 spec) is invoking the handler and *then* unwinding the stack if needed. 20:56
Tene I should be able to hack that up tonight. I could rebase my exceptions_refactor branch on top of head.
jnthn_ I think Parrot today unwinds the stack while looking for the handler.
The upshot is that resumable exceptions need not be continuations.
You just mark the exception object as resumable.
Tene Hmm. 20:57
jnthn_ And then it says "oh, then I guess I needn't unwind the stack"
The thing is, Parrot doesn't bound exception handlers in a code sense.
In what I'm working on, all handlers are blocks, never labels.
Going to see if I can get it like that. 20:58
*get it working
Tene So, if I have an exception handler in a block, and I call a function, and an exception is thrown somewhere in that other function, my exception handler isn't lexically in that scope, so how do I catch that exception?
jnthn_ Tene: S04 feels more concrete these days, fwiw. The "Exception handlers" section actually has code in it :-)
Tene orly?
jnthn_ The handler is in the dynamic scope of the thrower, but in the lexical scope of the catcher. 20:59
The point of invocation doesn't affect the outer, just the caller.
Hmm...I guess you can get into loops if you throw while handling. 21:00
Tene I'm sorry, I don't actually follow what you're looking for with "throw an exception lexically", then.
21:01 bluescreen left
jnthn_ Ah, there's two distinct things here. 21:01
The "throw lexically" means "look for a handler lexically"
Tene The labels bit in parrot exception handlers has always bothered me.
jnthn_ The actual invocation of the handler should look the same whether it was found dynamically or lexically.
Tene So you search lexically and dynamically? both? what order? 21:02
jnthn_ It's kinda the same as $foo() vs $*foo() semantics
We need a couple of ways to throw.
throw today just looks in the dynamic scope
We need a throw_lexically or something.
That looks for a handler through the outers. 21:03
wagle_ hi-- which command's documentation do i read to output the head1 NAME contents of a directory of .pl files?
jnthn_ But both of them, having found the handler, should invoke it the same way, so far as I can figure out. 21:04
Tene jnthn_: so, with this program, what's the decision procedure to work out which handlers to execute? gist.github.com/641368
jnthn_ (Again, speaking from half-written prototype.)
wagle_ (these are parrot files)
jnthn_ Tene: die always throws dynamically
Tene: We get that right today. 21:05
Tene: The one we get wrong is e.g. return
Tene Okay, what's the lexical exception handler for?
Ah.
I didn't understand that.
jnthn_ The canonical example is
sub foo(@x) { @x.map: { $_.foo || return } } 21:06
map is just a method so if we throw dynamically for return, then we end up hitting map's return handler first.
Tene Ah.
jnthn_ Whereas that's meant to return from the (enclosing) foo
Tene Yeah, I get it now.
Do blocks as exception handlers even work in parrot? I never saw that working. 21:07
jnthn_ I'm not sure. :|
Tene pir subs, I guess I should say.
Maybe I should start with that.
jnthn_ I'm not sure if they do.
Yes, that'd also be helpful
21:09 perlite_ joined
sorear jnthn_: I talked to TimToady about that case the other day 21:10
jnthn_: he said that control exception throws *without labels* are purely dynamic
which made my life much nicer :)
jnthn_ sorear: return has an implied label? 21:11
sorear return doesn't have a label
jnthn_ But that's...not dynamic?
sorear return is dynamic, it doesn't pay attention to lexical scopes at all
Tene jnthn_: fwiw, you *can* do that right with just dynamic. You just include a reference to the target sub in the exception, and have the exception handler's can_handle method check for that, and decline if it has a reference to a destination that's not the sub it's handling exceptions for. 21:12
jnthn_: I wrote a prototype of that that worked just fine, but seem to have misplaced it.
jnthn_ sorear: spec commit that changed that?
21:13 perlite left
sorear extrapolation from TimToady's word and a826b588 21:14
jnthn_ sorear: I think you've over-generalized that spec diff. 21:16
21:16 wagle_ is now known as wagle
jnthn_ sorear: It's talking about loop modifiers, not control exceptions generally. 21:16
sorear: Though if you want to be purely dynamic I guess you can implement return as tene++ suggested to get the right semantics. 21:17
sorear: Why did it make your life easier, ooc?
sorear: e.g. what am I going to run into that makes it harder than walking the outer chain rather than the caller one? 21:18
21:20 dukelet0 joined
Tene jnthn_: Well, I'm getting back to $realjob now. We'll see how I'm feeling when I get home. 21:21
jnthn_ Tene: kk. :-)
Enjoy your evening, whatever you end up doing. :-)
21:23 dngor_ joined 21:26 dngor left 21:28 whiteknight joined
tadzik wklej.org/id/405550/ -- is there anything wrong with this nqp code? I remember it working some time ago 21:37
oh nevermind, I ran it with parrot rather than parrot-nqp
wagle dukeleto: whats the other way to create skeletons (as opposed to create_language)? 21:38
21:49 dukelet0 left
dukeleto wagle: mk_language_shell.pl 22:01
wagle: in the same dir
wagle dukeleto: thanks
22:03 patspam left 22:05 jsut_ joined
wagle dukeleto: 22:08
wagle@comonad:~/research/scheme$ cat build/PARROT_REVISION
$Revision$
wagle@comonad:~/research/scheme$
from a pure git checkout an hour or so ago 22:09
22:10 jsut left
plobsing wagle: well sure. you expect it to be able to pull an svn revision out of a git checkout? that's like expecting apple juice from oranges. 22:11
wagle no, but its what dukeleto's create_language script does
oh well, gotta go for 2 hours.. 22:12
dukeleto wagle: what's up?
wagle the above file doesnt get the revision var expanded for create_language 22:13
you going to be at barcamp?
gotta go in 1 minute 22:14
dukeleto wagle: no, i am in CA for the GSoC Mentor Summit
wagle ok.. will explain better later
22:15 dngor joined 22:17 Andy left, dngor_ left 22:21 brianwisti left 22:32 kid51 joined 22:39 tadzik left
kid51 seen magnachef? 22:46
aloha Sorry, I haven't seen magnachef.
GeJ Bonjour everyone. 22:59
whiteknight hello 23:05
dukeleto hola
23:11 dngor_ joined 23:14 dngor left 23:36 dduncan joined 23:37 dduncan 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!