weekly Rakudo status meetings with phase transitions: Tue 19:00 UTC | IR clogs at irclog.perlgeek.de/phasers/today
Set by moderator on 20 September 2010.
sorear The worst trolls are the ones that I'm insufficiently certain about to just kick. 03:54
ok, it's now a language bashing fest... can I please start kicking people? 04:02
04:03 ash_ left 04:04 ash_ joined 04:26 ash_ left 08:17 Tene left 12:50 ash_ joined 14:22 ash_ left 14:30 ash_ joined 16:02 gottreu joined 18:34 mberends joined 18:55 masak joined
masak o/ 18:59
Util o/ 19:00
jnthn \\o
mberends \\o 19:01
masak my report: applied enums patch. \\o/ still ways to go on that one, though. also did minor things in IO and Instant. plan for next week: focus on the book. .eor
mberends masak++ 19:02
jnthn masak++ 19:03
Util My report: Wrote many Perl 6 solutions for RosettaCode. Attended first #phasers. .eor
masak Util++ 19:04
moritz_ very much occupied with real life; blogged a bit; started to write a Sudoku solver; small bits here and there.
masak oh, that reminds me: I now have a blog of my own. it's written in Perl 6! strangelyconsistent.org/
ash_ i made some changes to try.rakudo.org, tried to make the server a bit more stable, blogged about it too 19:05
mberends 6model/java is progressing well, no blockers, simply many more tuits required
jnthn mberends++ 19:07
masak mberends++
jnthn I has a report
Over the last week: 19:08
* Gave a talk in Oslo to NUUG. Seemed well received, and the video turned out really well too. sjn++ for organizing.
* Dug into the Parrot port of 6model. Got pure prototypes in place and able to do method dispatch already.
* Started to stub in more representations too
* Started to refactor handling of packages/meta-objects in nqp-rx on Parrot
This week I want to:
* Finish up porting the representations I'll most immediately want/need, especially P6opaque and it's slot calculation.
* Wrote the first in a series of explain-y blog posts.
* Get nqp-rx handling of knowhow syntax working. Nearly there.
* Get attribute access and binding working in 6model. Then port that over to the Parrot implementation.
* Start sketching out ClassHOW.
Blockers:
* Can't emit .loadlib properly yet using PAST -> PIR. Discussed a bit with pmichaud++ - mostly now just need a good name for the thingy.
</rapport>
pmichaud jnthn: I'll add that later tonight or early tomorrow.
jnthn pmichaud: Oh, outstanding. 19:09
pmichaud++
I was gearing up to do it myself. ;-)
masak jnthn++ # nuug talk
jnthn But that means I can focus on the reprs etc.
moritz_ jnthn++
mberends jnthn++ indeed 19:12
jnthn mberends: Catch me if you need help working out the stuff in dotnet/compiler/ 19:14
mberends: I readily admit some bits of it may be a tad obtuse. :-)
mberends jnthn: thanks. I'll add comments here and there to express my (mis)understanding and give others some more explanations. Your blog post related to the code fairly well :) 19:17
jnthn mberends: You made any decisions about whether to keep DNST or to copy it and make JST?
(e.g. keep it and write a DNST2Java.pm)
We can rename it to something more generic if so. 19:18
mberends jnthn: it's JST in name, but operationally identical where possible
jnthn mberends: I meant more the data structure than the code working on it. (more)
DNST.pm doesn't actually do anything other than specify a bunch of classes that hold stuff.
All the logic is in PAST2DNST.pm and DNST2CSharp.pm 19:19
mberends I renamed all occurrences of DN to keep a clear separation. We can merge whenever we feel like it, they're so close. 19:20
jnthn OK, if you're more comfortable to merge that one later, that's fine by me.
I hope we can later merge or find a way to share large parts of the PAST2DNST.pm too
e.g. maybe a common role or base class
But I can't guess what'd go in the role/base yet. :-) 19:21
mberends I expect the answers will only become clear after the Java port is complete and working 19:22
jnthn *nod*
Yes, it's too early to know yet.
mberends the ClassLoader system can also turn a raw array of bytes into a class, if the bytes conform to JVM standards. 19:23
jnthn *nod* 19:24
Yes, that'll be useful for the setting loading.
mberends jnthn: which files in 6model/dotnet are likely to change in the next week or two? 19:27
jnthn Actions.pm and PAST2DNST.pm 19:28
In compiler
In runtime, harder to say. 19:29
mberends ok, thanks
jnthn The multi-dispatch stuff is certainly going to get a good whacking at some point.
It kinda depends what I choose to do first.
I think the attributes stuff will mostly affect the compiler at this point. 19:30
But there's plnety more to prorotype
I may change some stuff in Ops.pm
er, Ops.cs
To work on pre-unboxed values rather than boxed ones.
And maybe finish up lexical lookup improvements. 19:31
There's so many bits to do. :-)
At some point P6opaque.cs is going to get a bit of a beating
But I may well try out the native typed attributes in the Parrot impl first.
Mostly because I have raw memory access and so can do disgust^Wbeautiful things. 19:32
mberends jnthn: some days, you should stick to C 19:33
jnthn Well, I guess it has to make *something* easier. :-)
mberends does MultiDispatcher.candidateSort() actually sort? Is List a sorted list? 19:36
jnthn Not really
It's a big pile of cheat at the moment
mberends lol 19:37
jnthn I wouldn't spend time to translate that to Java at the moment.
mberends it's already done :/
jnthn I've still got a bunch of stuff in my notebook that hasn't made it from diagrams to code yet.
OK
Oh, is_narrower is correct still I guess
So it's not all wasted effort
Actually looking at it, it's more incomplete than wrong
mberends no worries 19:38
jnthn The problem isn't the sorting, it's more the whole way we obtain/store the candidates to sort.
I'm kinda still exploring the design/implementation space there.
Not 100% sure how it should look.
moritz_ clean and lean, please 19:39
masak preferably perfect on the first try.
jnthn moritz_: Well, yes, they're requirements. But a washed beefsteak is clean and lean, so they're clearly not the only ones for a good multi-dispatcher. :P
masak .oO( I like my multi dispatchers like I like my beefsteaks... ) 19:41
I take it the meeting's over...? 19:52
jnthn I'm guessing so, unless pmichaud has anything to add. 19:53
mberends yeah, 11 minutes with no comment is idle enough
moritz_ like, a report?
pmichaud I didn't do a whole lot today.
er, this week.
let me check the logs.
masak :)
pmichaud fixed Any.reverse flattening 19:54
restored the -c option
fixed encoding of the standard filehandles
fixed the metaop forms of ?&, ?|, &&, etc.
fixed time/now to be properly 0-ary terms 19:55
fixed a nqp bug with negated zero-width character matches at the end of a string
eor
masak \\o/
pmichaud++
"didn't do a whole lot"... :P
moritz_ pmichaud++
jnthn "not much" :P 19:56
pmichaud++
pmichaud mostly I seem to be in recharge mode right now, as opposed to writing lots of code
jnthn pmichaud: It's still a fix/improvement per day of the week. :-)
pmichaud I've also been looking at parrot and rakudo performance improvements
jnthn is happy about today's one :-)
We'll have to see how it impacts Rakudo.
pmichaud for example, we made a change today (not in time for parrot release) that should speed up subroutine calls a fair bit.
moritz_ pmichaud: recharge mode is very understandable after the huge R* marathon earlier this yer
it's a bit similar here 19:57
jnthn +1. Important to get some rest.
pmichaud right
jnthn 'tis why it took me until Saturday to really dig into the Parrot porting of 6model
Though actually once I got going, it was a *lot* less painful than I expected.
Probably 'cus I, er, ignored most of Parrot... :-)
pmichaud \\o/ (less painful)++
moritz_ :-)
pmichaud were the nqp parts painful, ooc?
jnthn Well, all I've done in nqp is twiddle with the makefile and moving a few things into nqp-rx that I'd already got in JnthnNQP. :-) 19:58
pmichaud ah.
jnthn So not hit the Hard Bits yet in that sense.
pmichaud I'm still thinking about how/when I might want to switch PAST::Compiler to nqp source 19:59
jnthn Only pain point so far is loadlib.
Yeah
pmichaud maybe it'll become NAST :-)
then we just need a 'Y'
jnthn Ooh, NAST
It'll take me months to stop typing PAST... 20:00
After 2+ years or doing so week in week out... :-)
pmichaud :)
it's likely to remain PAST, somehow.
jnthn Perl6-ish Abstract Syntax Tree
pmichaud anyway, that's all I have for this week.
moritz_ Portable Abstract Syntax Tree 20:01
pmichaud for this upcoming week I'll be helping with compiler/star releases and researching cpan-ish things.
jnthn +1
pmichaud: And if you manage it, the loadlib unblocker. :-)
Or just tell me what to call it. :-)
I guess that's $meeting.end 20:02
Thanks, all
.oO( To the stejkhus! )
masak :D 20:03
20:05 masak left 20:10 ash_ left 20:38 mberends left 23:25 ash_ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!