Parrot 4.2.0 "Ornithopter" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 21 March 2012.
whiteknight NotFound: That last example in my gist still doesn't work 00:26
gist.github.com/2159106#file_generated.pir 00:27
the .const 'Sub' line happens after the optional args are set up. I'll take a look at it
msg NotFound: github.com/Whiteknight/winxed/comm...f29f150be4 00:41
aloha OK. I'll deliver the message.
00:49 benabik joined 01:01 bacek_at_work joined, jashwanth joined 01:02 kid51 joined
bacek_at_work aloha, humans 01:02
01:07 alvis` joined
whiteknight hello bacek 01:09
bacek: Excellent branchwork
bacek_at_work whiteknight, thanks. Now stop slacking and jump on this branch to finish it :) 01:10
whiteknight bacek_at_work: does this branch build? 01:19
bacek_at_work whiteknight, 01:20
whiteknight, yes. But new Signature isn't hooked up at all.
whiteknight src/call/signature.c:66:3: error: redefinition of typedef 'Parrot_Signature' is invalid in C [-Wtypedef-redefinition]
wtf, that error doesn't make any sense 01:21
nevermind, I'll make it work 01:23
bacek_at_work whiteknight, good, good. Come to the dark side, young padavan. 01:24
whiteknight compilers/pge/PGE/builtins.pg appears to hang 01:25
bacek_at_work whiteknight, just use make corevm. I had few changes which blocks full build. 01:26
whiteknight okay
I need to review your changes more closely before I can do any work
kid51 msg alester [ot] Your debugging expertise is cited in rt.perl.org/rt3/Ticket/Display.html?id=71678 01:27
aloha OK. I'll deliver the message.
bacek_at_work aloha, seen dalek 01:40
aloha bacek_at_work: dalek was last seen in #parrot 13 hours 41 mins ago saying "parrot/pcc_reorder: review: github.com/parrot/parrot/commit/d3630a9ae9".
bacek_at_work meh....
whiteknight, github.com/parrot/parrot/commit/93...518fbbfdb2 01:41
this should give overall idea how to proceed :)
01:45 benabik joined
jashwanth hello whiteknight 01:46
whiteknight hello jashwanth 01:50
jashwanth: where do you live?
alester kid51: Not exactly. He's just referring to a book on the debugger that I have a co-author credit on. 01:58
kid51 alester: Well, if you have any thoughts on those issues, feel free to post. 02:04
kid51 has never used 'a' command in debugger. 02:05
alester Oh hell no. I don't know the Perl debugger at all. :-)
kid51 How did you co-author the book, then?
alester My name is a co-author because I helped Richard drive it to completion and did a lot of edits.
kid51 Ah, you cracked the whip!
alester Kinda. They were about to kill the book, and I said I'd help 'im finish it. 02:06
awwaiid fancy 02:07
02:13 mdupont joined 02:57 nbrown joined 03:22 GeJ joined 04:45 fperrad joined 04:48 GeJ joined 05:19 preflex_ joined 07:12 bacek_ joined 07:18 mj41 joined 07:21 TonyC joined 07:22 nopaste joined 08:16 lucian_ joined 08:17 cosimo_ joined 10:24 bacek_ joined 10:25 bacek_ joined 10:26 bacek_ joined 10:43 bacek_ joined 10:46 bacek_ joined 10:56 fperrad joined 12:16 whiteknight joined
whiteknight good morning, #parrot 12:16
tadzik hello whiteknight 12:20
whiteknight good morning, tadzik 12:22
Infinoid sorear: pong 12:35
whiteknight Infinoid! 12:40
Infinoid hi whiteknight :)
whiteknight Infinoid: How are things? 12:41
Infinoid Still busy, but good 12:44
I made the mistake of getting hired by a company that makes me actually use my brain... so I'm not left with a lot of bandwidth for other things 12:45
Sorry for disappearing off the end of the earth
moritz if it makes you happy, it was the right choice :-)
whiteknight Infinoid: what company are you at now? 12:50
or is it the same place? 12:51
Infinoid Same place. www.etinternational.com/
whiteknight cool 12:52
masak Infinoid! \\o/ 12:55
12:58 bacek_ joined 13:47 hercynium joined 13:48 JimmyZ joined
JimmyZ hello #parrot 13:48
looks like m0 is deviated from trac.parrot.org/parrot/wiki/LoritoRoadmap 13:50
13:55 PacoAir joined
whiteknight hello JimmyZ 13:56
13:57 JimmyZ_ joined 14:13 alester joined 14:25 jashwanth joined 14:29 lateau joined
NotFound whiteknight: ping 14:31
14:31 PacoAir joined
whiteknight NotFound: pong 14:31
NotFound whiteknight: What does that patch? Emiting const subs before patameter defaults? 14:32
whiteknight function foo(var x = NameSpace.func()) { } 14:33
that doesn't work before the patch
NotFound Ah, yes, forgot to add that part in my test program.
whiteknight doesn't look like anything else breaks
NotFound whiteknight: Good, going to merge it. 14:34
whiteknight thanks
NotFound++
NotFound whiteknight: about Guitor and gsoc: low level xlib is hard, I doubt any student will pick it.
14:35 benabik joined
whiteknight NotFound: if they pick it or not, we can still write the idea 14:35
NotFound: you picked xlib. Somebody else might be like you :)
benabik Good morning, #parrot
whiteknight hello benabik 14:36
14:36 PacoAir joined
NotFound whiteknight: yes, but these days people tend to use Cairo or the like instead of dealing directly with X. 14:36
whiteknight I would love to have Cairo bindings too 14:37
NotFound That's a good idea.
I'd like a java copiler targeting parrot, or even a tiny java subset. 14:39
whiteknight yes, that was proposed last year
14:39 PacoAir joined
NotFound Yeah, but last year I hadn't any interesting java code. Now I have my new Basic interpreter. 14:40
benabik I would personally like to see more infrastructural work. Parrot needs low-level work to move forward.
But I'm a systems kinda guy. :-D
whiteknight benabik: you could change to a different project :)
moritz if we get Cairo bindings, who will maintain them?
benabik whiteknight: Well, I consider PCT/PIR to be low-level. :-D
NotFound moritz: they don't need to be in core, an external project is fine. 14:41
whiteknight JIT and Threads are the next two things that we really need to be added in core, and nine has Threads mostly done
All other changes in core are rewrites, cleanups, refactors, optimizations, etc
It's hard to say "I'm going to add X to parrot core", because there aren't a lot of well-defined X values that we need from the ground-up 14:42
benabik whiteknight: 6model, although that should really go along with a PMC refactor (with fire, perhaps)
whiteknight yeah, and that's something that we "have", if you don't mind dynloading it
NotFound Threading will allow interesting applications. 14:43
whiteknight yeah, I guess we can say "new object model", with the stipulation that 6model be a core part of it
NotFound: yes, I'm very excited. 14:44
NotFound Now that I think about it, there is a thing that can be interesting for gsoc: improving NCI callback capabilities. 14:45
moritz or make NCI not dependent on a library that's hard to build on windows 14:46
NotFound I need a way to set callbacks for Guitor to handle X errors.
whiteknight yes, a new callback system that used the power of the new threads would be awesome
if our scheduler is good, it can pause the current thread, execute the callback, then resume the thread 14:47
NotFound Yes, but the hard part is doing callbacks without PMC-able parameters.
whiteknight you'd have to wrap them in something 15:07
15:20 dmalcolm joined 15:25 rich joined
rich hey i'm playing around with PCT and I dont understand what || should mean in this context: [ $ || <.panic: "Syntax error"> ] 15:26
is it double | ?
benabik It's alternation. 15:27
In p6 rules, | matches whichever is longest and || tries each one in turn.
rich ok thx 15:28
benabik I'm not sure nqp-rx actually implements that, but that's the syntax.
moritz correct
rich what about the python aka pynie for parrot? it seems it is orphaned? 15:29
moritz it is
whiteknight rich: there were a few interested developers, but it never reached a self-sustaining critical mass 15:57
NotFound The funny thing is that critical mass seems to be around 1. 16:02
rich :D 16:03
so how complete is it? 16:04
NotFound Blaming mode on. I mean, usually some people say "I'm very interested in X", but don't have enough interest on X to keep looking on it even for a few monts. 16:05
benabik Critical mass is about 1, but very few people provide an entire person. 16:08
moritz looking isn't the problem. Working on it is :-) 16:10
NotFound Yeah, but if they don't even look... 16:15
whiteknight rich: I have no idea, to be honest
rich: a big part of the shortcoming of pynie was the object model of it, which doesn't quite match python semantics and was holding back implementation of any kind of standard library 16:16
dukeleto ~~ 16:18
rich hm, i'm not that experienced building compiler... can you explain that in a little more detail?
dukeleto rich: there is an example language called squaak
rich well i did that tutorial
and also we have some lectures in univercity 16:19
dukeleto rich: ok.
benabik rich: Parrot's idea of an object is slightly different than Python's idea of an object. The incompatibilities made it difficult to work on building the normal Python libraries.
dukeleto rich: what part do you have questions about? which language are you using?
rich: right now, people can use Winxed, PIR and various flavors of NQP to write a high level language (HLL) 16:20
rich: or you can roll everything yourself :) But that isn't for the faint of heart.
rich dukeleto: hm well i was thinking to create a subset of python or something similar (cause i like the lang) to get myself confortable with parrot. cause pynie looks quite orphaned i was wondering why 16:21
16:21 jsut_ joined
rich benabik: do you think to get the stdlib working with parrot? 16:22
or is it just not worth the effort?
whiteknight rich: do you happen to be a college student? 16:23
sorear hey Infinoid
benabik rich: I think it should be possible to get Python objects working on Parrot. Not simple though. One of the reasons I keep bringing up putting 6model in core is because it would simplify things immensely.
rich whiteknight: y
whiteknight rich: might you be interested in working on a python compiler for Parrot for GSOC this summer? 16:24
NotFound whiteknight: You haven't made a pull request for that patch?
whiteknight NotFound: not yet
NotFound: I will now
NotFound Ah, I misunderstood it was one.
dukeleto rich: have you seen puffin? That is a newer relative of pynie 16:25
rich: but still, both are blocked by a better meta object protocol being in parrot core
rich whiteknight: hm sounds good
dukeleto rich: so the plan is to borrow 6model from the perl 6 core, but 6model is still a moving target... 16:26
rich dukeleto: not yet ill check it out
whiteknight NotFound: Can't make the pull request, I have another garbage commit in there that I need to remove first
dukeleto: It's not moving so far so fast
NotFound A doubt: Why they need it to be in core? Can't they load 6model?
whiteknight NotFound: Yes, we can do that
benabik NotFound: Incompatibilities between 6model world and PMC land.
whiteknight NotFound: A python compiler probably should do that, at this point 16:27
NotFound whiteknight: NP, I'll get the patch.
whiteknight NotFound++
NotFound benabik: Such incompatibilities doesn't seem to block rakudo. 16:28
benabik NotFound: Because rakudo only works with rakudo-land. I'd like to at least try to have a world where our languages can work together.
NotFound: Things can go slightly sideways in Rakudo if a raw PMC gets handed back to the "user". 16:29
NotFound is tempted to steal 6model code and call it funmodel 16:30
whiteknight benabik: That's Rakudo. They use all sorts of object context and metaobjects to control things. A Python compiler wouldn't make any of those assumptions
dukeleto Github v1 and v2 API are going away soon: github.com/blog/1090-github-api-moving-on
will that affect any of our bots?
NotFound Or even better, call it topmodel 16:31
dukeleto NotFound: funmodel++
benabik Really, in the end, I think that 6model is a more flexible, more useful object model. So rather than hang onto PMCs, it would be better for Parrot in the long run to adopt it. 16:33
Also, I'm relatively certain that we could adapt the existing PMC model to work within it for compatability.
benabik sighs.
Now I'm tempted to do that for GSoC. But I wanted to work on my shiny. 16:34
whiteknight benabik: There's nobody saying you can't do both...
benabik whiteknight: I think thesis, baby, PACT, and 6model would be a bit too much.
whiteknight benabik: if you had asked me a few months ago, I would have told you to hold off on the baby :) 16:37
benabik whiteknight: :-P 16:38
NotFound Hell... there wasn't a way to get a commit's diff from gihib? 16:39
whiteknight I could have taken a dose of that medicine myself. I don't think anybody is ever quite prepared for it
NotFound: I think it's x..y in the url 16:40
dukeleto benabik: we need a transition plan where PMCs boil down to "funmodel" or whatever parrot core will call it 16:41
NotFound: of course there is
benabik dukeleto: Import 6model. Create a PMC_HOW and PMC_REPR that forwards the 6model bits to the appropriate PMC bits. Migrate our defaults from PMC to 6model. 16:42
benabik sees little point in changing the name. 16:43
whiteknight benabik: yeah, that's my general plan for it
I want the import 6model into core, in parallel with our existing model, then start slowly eating away at the old version
benabik It doesn't seem terribly difficult to me, but there is a fair amount of "SMOP" in there. 16:44
whiteknight every now and then I'll enter into a fit of uncontrollable rage and delete a whole file of old code
dukeleto benabik: sounds like a good gsoc proposal .... 16:45
NotFound: github.com/parrot/parrot/branches
NotFound: there is a "compare" button on the righthand side of all the branch names 16:46
Does anybody have a problem with me merging this branch? github.com/parrot/parrot/compare/m...%2Fdlclose
It works fine on Linux, and it isn't getting tested enough on the lonely branch. It should remove the need for a custom patch for the netbsd people 16:47
whiteknight dukeleto: let me look
dukeleto: looks swell to me
dukeleto whiteknight: hokey dokey
NotFound Note that in some platforms (hp-ux for example) dlclosing is not refocounted. Don't know if we support such. 16:48
16:49 contingencyplan joined, bacek_ joined
NotFound For example, you can nicely kill the program by dlclosing C std. 16:49
dukeleto NotFound: not sure we support HPUX 16:51
NotFound: and if an HPUX user sends a patch, I will gladly code review it :)
NotFound dukeleto: I think it was ten years ago the last time I used hpux.
Got it! It was just adding ".patch" to the commit url. 16:53
dukeleto NotFound: ah, yes. You can add .diff too
benabik .patch includes the commit message, I think. 16:55
dukeleto NotFound: yes, .patch is the actual git patch with author info, .diff is just a diff that can be given to straight up "patch" 17:02
NotFound whiteknight: applied, tested, and snapshot updated 17:17
whiteknight NotFound++
17:50 whiteknight joined
whiteknight I sent a message about GSOC to some of my old professors, to pass along to interested students 17:50
They clearly didn't read the FAQ, and seemed to fundamentally misunderstand what the program is about
so now I'm getting emailed things that are essentially grant applications from students for their academic work 17:51
...which is kind of interesting to see what kinds of projects students are working on over there, but not exactly what GSOC is all about 17:52
I may have to just go over there one day and give an in-person presentation about it, to prevent misunderstandings 17:59
PerlJam whiteknight: that's okay, I send info about gsoc to the local csclub every year and so far I've gotten approximately 1 response of "what's this?" and that's about it.
whiteknight I don't understand how more people don't get more excited about this kind of opportunity 18:00
it's like a paid internship, where you work from your house on a real-world software development team
PerlJam too many unknowns 18:02
whiteknight if I were still in college, I would be jumping up and down like a coked-up chihuahua
tadzik most of the people I've told do act like this :) 18:03
mostly due to OMG GOOGLE MONEY FOR SITTING AT HOME
PerlJam If there's not already involves with an open source project, they have to jump the hurdle of participation in something they currently know nothing about.
s/there's/they're/
*involved
NotFound Thinking about some people I frequented years ago, probably many think they are going to steal his incredible powerful, original and high valuable ideas. 18:04
I sometimes sees some like that on irc. They ask naive doubts about trivial programming tasks, and refuse to show the code because we can steal it! 18:05
whiteknight I've gotten lots of emails from students so far. I haven't received too many follow-ups yet though
PerlJam NotFound: insane people aren't useful anyway :) 18:07
NotFound PerlJam: they are useful! They make me smile in boring days :) 18:08
nine CORE.setting++ # gives me much time to catch up with the day's chatter 18:17
tadzik :) 18:18
nine whiteknight: removing orig_interp from the PMC header gets me back to 2.47% slowdown compared to master with signficance 7.9Ļƒ 18:26
whiteknight nine: Yeah, we know we have to find a better way to do that
nine whiteknight: I recycled an interp flag to tell the GC to flag newly created PMCs while doing a proxied call. 18:27
whiteknight: so apart from being very convenient for my assertions, I don't need the orig_interp anymore 18:28
whiteknight nine: okay, that's cool
okay, add it under an #ifdef THREAD_DEBUG block
or #ifndef NDEBUG 18:29
nine whiteknight: still not exactly sure where the slowdown is coming from. Will do a new benchmark run on master to see if it's just some fluke in my benchmarking setup
whiteknight so that you can have it when you want it, and not in an optimized deployment
jashwanth dukeleto:hello 18:31
whiteknight dukeleto: ping 18:37
msg dukeleto jashwanth is the prospective GSOC student I was telling you about who wants to work on PLA. You two should be introduced
aloha OK. I'll deliver the message.
whiteknight msg jashwanth aloha is a bot who delivers messages 18:38
aloha OK. I'll deliver the message.
dukeleto whiteknight: pong 18:46
jashwanth: howdy
NotFound Talking about bots, what happened to dalek? 18:47
whiteknight NotFound: off to the great electromagnet in the sky
Actually, I have no idea. Where is that damned bot
benabik It went missing yesterday. Haven't seen a reason yet.
nine it quit after a series of commits by bacek++ 18:48
whiteknight it has a lot of trouble with large lists of commits
tadzik there's no room for two robots on a channel it'd seem
whiteknight One of the next projects I want to work on is an IRC client library for making bots 18:49
then we can write all our infrastructure droids in pure parrot 18:50
18:51 rich left
whiteknight I really want a bot that has commands to update and recompile its own sourcecode 18:51
We're going to need to majorly redo namespaces before that is possible 18:52
and maybe security sandboxing 18:53
Oooh! That would make a great GSOC project!
What I really need to do is keep track of project ideas so we can reuse the good ones next year 19:31
benabik whiteknight: Isn't that what the wiki is for? :-D 19:32
whiteknight well, yes
but...shaddap
I also need to keep better track of which projects get done each year. We don't have anything like a hall of fame with information about completed projects 19:33
dukeleto whiteknight: sounds like a nice page for parrot.github.com 19:35
nine whiteknight: seems like the benchmark somehow sucks. I get too high variation for conclusive results. 19:56
whiteknight nine: Okay, so assume a bias in your favor and call it a victory 20:01
moritz the mandelbrot benchmark? 20:02
20:05 jsut joined
Coke I forwarded whiteknight's blogpost to RPI's open source maven. 20:23
(re GSOC 2012)
whiteknight which blogpost? 20:28
actually, I'll find out later. I have to go home now. 20:31
Coke "we're in" 20:35
bacek_at_work ~~ 22:55
aloha, humans
seen dalek
aloha dalek was last seen in #parrot 1 days 10 hours ago saying "parrot/pcc_reorder: review: github.com/parrot/parrot/commit/d3630a9ae9".
bacek_at_work Did I kill dalek?
dukeleto bacek_at_work: yes 23:00
bacek_at_work dukeleto, sigh...
How hard is to implement github commits announce in aloha? Mmmm... Tempting 23:01
23:06 whiteknight joined
whiteknight good evening, #parrot 23:10
23:48 perlite_ joined
dukeleto whiteknight: wazzup 23:50
whiteknight: what is your take on the gsoc student email on parrot-dev? It needs a lot more beef.
whiteknight: do you have a specific PLA task that needs doing? 23:51
whiteknight: also, can you maybe give parrot-dev a "state of the PLA" ? It hasn't seen much love lately and I am unsure what still needs doing.

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!