»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:04 dorlamm joined
diakopter finds it funny that the syntax highlighting setting that works the best for Actions.nqp is the C language setting. 00:07
sorear I should try that 00:08
I've been using filetype=perl for most of my Perl 6 hacking
arnsholt Not petdance/vim-perl?
diakopter this is in Notepad++, so it probably won't match up with vim 00:09
sorear which works pretty well but has an unfortunate tendency to get stuck in the wrong mode
arnsholt Yeah, that makes sense =)
Yeah, it does get lost occasionally
sorear arnsholt: the problem with petdance's perl6 mode is that it is infuriatingly slow
at 300-500 lines I can type faster than vim can update the highlighting (and it does so synchronously, so my typing is slowed down) 00:11
at 2000-6000 lines just _opening_ files takes far too long
this is of course not a complaint against petdance. Rather it is an instantiation of Alias' old complaint that Perl 6 cannot be effectively parsed by tools 00:12
diakopter should m:Perl5// support codeblocks? 00:13
sorear why not?
diakopter ok
00:14 tokuhirom left
diakopter sorear: should m:Perl5// be sigspace? 00:17
00:19 ksi left 00:20 Sarten-X joined
sorear diakopter: I do not understand why you are asking this question. Yes, spaces are significant in Perl 5 regexes 00:21
00:21 replore joined
diakopter ok, I was thinking of //x 00:21
arnsholt sorear: Yeah, those are good points 00:25
Most of my Perl 6 is smaller than that though
00:44 cooper left 00:45 cooper joined 00:51 mdxi joined 00:57 PacoLinux joined 00:58 Trashlord left 01:00 Trashlord joined 01:04 PacoLinux left, Chillance left 01:11 woosley joined 01:14 lestrrat joined 01:23 dorlamm left 01:27 Mowah_ left 01:32 PacoLinux joined 01:38 PacoLinux left
diakopter making <[\c32 .. \c126]> work is quite tricky 02:11
esp if <[\c1 .. Z]> is allowed 02:12
niecza: say '"' ~~ /<[\c32 .. \c126]>/ 02:15
p6eval niecza v11-24-gcdd4e66: OUTPUT«===SORRY!===␤␤Action method charspec not yet implemented at /tmp/L7CMFR3epG line 1:␤------> say '"' ~~ /<[\c32⏏ .. \c126]>/␤␤Action method backslash:c not yet implemented at /tmp/L7CMFR3epG line 1:␤------> say '"' …
02:28 cognominal_ joined 02:33 cognominal_ left, cognominal___ joined, alvis joined 03:06 ggoebel joined 03:18 pat_js joined 03:21 envi_ joined, wolfman2000 joined 03:41 satyavvd joined 03:54 envi_ left
diakopter karma larry 04:18
aloha larry has karma of 157.
diakopter karma sorear
aloha sorear has karma of 2027. 04:19
04:19 Sarten-X left
sorear karma TimToady 04:20
aloha TimToady has karma of 250.
sorear diakopter: note that these figures are skewed because we've gone through several karma bots 04:21
diakopter yeah
karma audreyt
aloha audreyt has karma of 3.
04:22 PacoLinux joined
pat_js hi, does someone know which of these dispatch:<…> methods do the "normal" dispatch? 04:22
04:22 molaf joined
pat_js like I can override dispatch:<.=> to get something else for $foo.="whatever"() 04:23
but which of these do I override for a "normal" method?
perl6: my $foo = "mumble" but role {method dispatch:<.=> ($var){say "Got $var"}}; $foo.='blabla'() 04:24
p6eval niecza v11-24-gcdd4e66: OUTPUT«Potential difficulties:␤ Useless use of quotes at /tmp/ZLr6GROAcj line 1:␤------> ($var){say "Got $var"}}; $foo.='blabla'⏏()␤␤Unhandled exception: Unable to resolve method blabla in class Str but ANON␤ at /tmp/ZLr6GROAcj line 1 (mainl… 04:25
..rakudo 4cff8e: OUTPUT«Got blabla␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "$foo"␤ expecting "=", "::", "handles", "is", bare trait, subroutine parameters, trait, block, context, ":" or "("␤ at /tmp/a0zlS8XBeT line 1, column 4␤»
04:26 birdwindupbird joined
sorear pat_js: the dispatch:<.=> methods are, as far as I can tell, an undocumented and nonstandardized Rakudo implementation detail 04:26
pat_js ah, ok thanks 04:27
sorear pat_js: I suppose you are trying to do something like perl 5's AUTOLOAD? 04:28
pat_js sorear: sortof, just looking at Rosettacode to solve some simple task, and thought "Send_an_unknown_method_call" would be possible with all that metamodel stuff 04:29
sorear niecza: class Foo { method FALLBACK($name, *@args) { say "got $name" } }; my $foo = Foo.new; say $foo.defined; $foo.blabla # here's what niecza can do for you
p6eval niecza v11-24-gcdd4e66: OUTPUT«Potential difficulties:␤ @args is declared but not used at /tmp/DuccPvbalS line 1:␤------> class Foo { method FALLBACK($name, *⏏@args) { say "got $name" } }; my $foo = ␤␤Bool::True␤got blabla␤»
sorear pat_js: the "correct" Rakudo way involves overriding find_method on the metaclass 04:30
jnthn please correct me if I'm wrong - I don't think users are supposed to touch dispatch:*
sorear queries p6eval to try and get somthing working
patch hi all
i just fixed lots of typos in the specs 04:31
and am about to push them
should i update Last Modified or just leave it since these are not substantial changes?
diakopter when fixing typos I usually don't update it 04:32
it's up to you though
patch ok, thanks
04:37 PacoLinux left
dalek ecs: 45ca8c8 | (Nick Patch)++ | S (16 files):
fix typos
 04:43
diakopter patch: wow nice 04:44
04:57 jaldhar joined
pat_js sorear: and HOW do i override a method in the metaclass, I can't even access it properly (e.g. has no .clone) 04:59
sorear pat_js: jnthn blogged an example a few months ago 05:02
05:03 domidumont joined
sorear the spec talks about a CANDO method; Niecza's FALLBACK is closer but not quite there 05:04
05:07 molaf left
pat_js sorear: yes, I found it, but when I try to run his code, it says 'can only get_how on a SixModelObject' 05:27
sorear: sorry, made some copy & paste errors 05:28
05:36 pat_js left 06:01 localhost joined
diakopter sorear: how goes the Unicode strivings? 06:22
sorear diakopter: blocked on tuits 06:49
06:58 koban joined 07:04 sayu joined 07:06 wtw joined 07:15 kaleem joined 07:23 wtw left 07:35 wtw joined 07:48 wtw left 07:50 wtw joined 08:16 koban left 08:24 mkramer joined, mkramer left 08:26 y3llow_ joined 08:28 y3llow_ is now known as y3llow, Lothar joined 08:31 mdxi left, mdxi joined, mkramer1 joined 08:35 spine joined, spine left 08:39 am0c joined 08:46 mj41 joined 08:50 PacoLinux joined
masak morning, #perl6 08:54
masak yawns
arnsholt G'day masak o/
08:54 PacoLinux left
masak wow -- I know I'm busy when it doesn't even occur to me to backlog... 08:54
sorear: sorry I missed your highlight yesterday -- I'd be happy to talk about promoting Niecza. 08:59
sorear heh.. now *I'm* busy 09:10
09:10 sftp joined, sftp_ left, jevin left, miso2217__ joined, Moukeddar joined, jevin_ joined 09:11 robins joined 09:18 Moukeddar left 09:19 mj41 left 09:20 sayu left 09:26 mj41 joined 09:33 icwiener joined, Sarten-X joined 09:35 mkramer1 left 09:36 mkramer joined
jnthn mornin' 09:36
09:36 woosley left
sorear heya jnthn 09:36
jnthn dang, slept through alarm... 09:37
jnthn is meant to be catching up $dayjob project today...
09:37 Sarten-X left 09:38 cosimo joined
jnthn sorear: The dispatch:<blah> are an implementation details. 09:38
sorear: I can hide them if it's problematic that they're there. 09:39
...somehow... :)
09:40 pomysl joined, pomysl left, pomysl joined
sorear jnthn: don't bother I say 09:42
jnthn: maybe rename them to _dispatch:<blah>
09:43 odoacre joined 09:44 odoacre left
jnthn sorear: Thinking about it a little more, they may end up as privates; those are non-vritual and so the optimizer would have a better shot at inlining them. 09:44
09:53 envi_ joined 09:55 odoacre joined 10:02 fsergot left 10:08 cosimo left
tadzik 'morning 10:12
10:14 bacek joined
bacek ~~ 10:15
10:15 replore left
moritz \o 10:15
bacek moritz, o/
jnthn bacek! 10:18
o/
10:20 cosimo joined
bacek jnthn, aloha :) 10:20
jnthn, I was looking for you, actually. 10:21
I do want to get rid of Class/Object/PMCProxy/Namespace stuff in parrot in favour of 6model.
10:21 fsergot joined, mkramer left
djanatyn awwww :( 10:21
there are no perl6 google code in tasks 10:22
I think
bacek jnthn, is github.com/jnthn/6model/blob/maste...erview.pod still a good starting point to understand 6model design?
10:32 daxim joined
sorear sleep 10:33
jnthn bacek: sorry, was on phone
bacek: Yes 10:34
bacek: The key header file to read is sixmodelobject.h
bacek: 6guts.wordpress.com/2010/10/15/slid...ymorphism/ is also interesting perhaps. Plus there's some other posts on that blog that may help 10:36
bacek jnthn, got it. Need couple of days to wrap my head around how it should fit into parrot. 10:43
jnthn, I'll (probably) ll have more questions later :)
10:43 cosimo left
jnthn bacek: That's fine :) 10:46
10:49 PacoLinux joined 10:53 PacoLinux left 11:00 amkrankruleuen left 11:01 amkrankruleuen joined, icwiener_ joined 11:02 icwiener left 11:07 envi_ left 11:10 icwiener_ left, kensanata joined 11:11 sftp left, y3llow left
mls morning! 11:12
fsergot mls o/
tadzik hello mls!
mls IMHO one thing that's missing in current 6model is a "linking" phase 11:13
Currently everything cas to be rebuilt when nqp is recompiled 11:14
I think this has to go in the future. Say a distribution ships with a lot of perl6 modules, all precompiled to pbc. Then there's a securtiy issue in perl6, su an update is released. 11:16
So I think some kind of "run time linker" at load time would be nice, which resolves the SC slots
(This also means that SC entries must be "named", so there can be a lookup of some kind) 11:17
jnthn mls: People should not be shipping PBCs ever 11:18
mls: I see the PIR/PBC as a cache
mls So you want to write them in the users home directory?
I don't think that makes sense
jnthn mls: Or the module installation tool is responsible for installing those. 11:20
mls yeah, but what happens if there's a perl6 update?
You need to "recompile" all modules as well
jnthn Correct.
daxim use $XDG_CACHE_HOME, which is $HOME/.cache by default
standards.freedesktop.org/basedir-s...atest.html
jnthn daxim: ooh, thanks
mls (except when there's no desktop) 11:21
jnthn (or you're on Windows ;))
daxim nonsense
it's also good on systems that do not have a desktop
I want to configure the backup regime to exclude cache dirs
mls oh, where is it set? 11:22
11:23 am0c left
daxim /etc/xdg and ~/.config/user-dirs.dirs, but really read the spec. it's not long. 11:23
mls is it set by some pam module at login time?
jnthn mls: I've already lived through the "you can update Rakudo without recompiling modules against it" days. People got a bunch of weird problems because things changed in subtle ways. 11:24
daxim called from /etc/profile.d
11:24 araujo joined
jnthn And there wasn't a clear error, stuff just...randomly broke. 11:24
mls Hmm, I have XDG_CONFIG_DIRS and XDG_DATA_DIRS, but to XDG_CACHE_HOME. Seems I need to update my system. 11:27
jnthn: But it's pretty much the same with "C" and shared libraries. 11:28
Saying you mist recompile everythink if there's a glibc update doesn't seem right
must 11:29
daxim: thanks for the xdb links, by the way. daxim++
xdg
one learns something everyday, I never heard of that spec 11:30
(oh, a "lennart" spec, is that a good or a bad sign ;) ) 11:31
daxim usually a bad thing
mls ;) 11:32
jnthn mls: Remember that Perl 5 doesn't really compile anything in advance; it's done each time. We've long thought of any pre-compilation as a kind of cache, with the source file always being the authoritative thing.
mls: Compare that to C, where you never install the sources.
mls: Well, unless you're in a development situation. 11:33
daxim grunts vaguely about statting .pmc files
mls Well, I compare it to python, which has that pbo files
(and we want to be faster than python ;) ) 11:34
jnthn mls: And not having to do a bunch of name resolution on startup will be faster ;) 11:35
mls yes, that's also true ;) 11:36
afk -> lunch
11:38 satyavvd left, mkramer joined 11:54 domidumont left
bbkr nom: say 123 ~~ int32 11:54
p6eval nom 4cff8e: OUTPUT«Bool::False␤»
11:57 domidumont joined, PacoLinux joined 12:10 mj41 left 12:12 Psyche^ joined 12:16 Psyche^ is now known as Patterner 12:22 replore_ joined 12:25 mj41 joined 12:28 replore_ left 12:30 PacoLinux left 12:45 am0c joined
dalek p: 13508b4 | mls++ | src/NQP/SymbolTable.pm:
use direct lexinfo access in install_lexical_symbol()
 12:49
12:51 sftp joined 12:55 am0c left 13:08 PacoLinux joined 13:10 PZt left 13:12 am0c joined, PacoLinux left 13:16 kaleem left 13:18 aloha joined 13:22 kensanata left 14:01 mkramer left 14:02 mkramer joined, mkramer left 14:03 mtk joined 14:06 molaf joined
[Coke] (perl 6 GCI tasks) we can rectify that at the second task-update point. 14:08
masak when is that? 14:09
14:10 mkramer joined, plutoid joined 14:12 mkramer left
[Coke] December 16, seems. 14:13
I'll ask in #gci on perl.org about that.
yup. no worries, we can get p6 stuff in for that round. 14:16
14:20 Trashlord left 14:21 mkramer joined, mkramer left
masak \o/ 14:24
14:26 plutoid left, envi_ joined 14:27 bluescreen10 joined
masak nom: say DateTime.new() before DateTime.new(:year(2011), :month(2), :day(28)) 14:28
p6eval nom 4cff8e: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:614␤ in sub infix:<==> at src/gen/CORE.setting:2024␤ in sub infix:<cmp> at src/gen/CORE.setting:1059␤ in sub infix:<befo…
masak since one can map DateTimes to Instants it could make sense to compare DateTimes... I think.
maybe there's something funny with timezones that I'm not realizing right now. 14:29
hm, seems the spec mandates a timezone. so it should be fine. 14:30
[Coke] We should probably get a list of tasks going ahead of time so we don't have a last minute rush. 14:31
masak speaking of last minute rush... 14:32
...allow me to gently insert the idea of a "Perl 6 Advent Calendar 2011" into the minds of gentlepeople on the channel.
nothing says we *have* to have one this year. but it's generally a good PR thing, and one that people appreciate. 14:33
PerlJam says we have to have one :)
masak administration is pretty minimal; we generally make do with one or two files in a repo somewhere to coordinate themes and authors. 14:34
PerlJam (of course, if I do all the entries it'll be very boring, but so be it)
masak I volunteer for 4 or 5 entries, evenly spaced.
jnthn I can do one about Grammar::Debugger, and probably one on meta-programming 14:43
14:43 PacoLinux joined
jnthn (if those are considered interesting topics) 14:43
PerlJam jnthn: aye
masak absolutely. 14:45
I bet there are plenty of subjects that we've covered in previous years but that we could cover again, slightly differently. 14:46
for example, there's probably lots more to write about junctions. or metaoperators. 14:47
jnthn Potentially I can do one on Zavolaj 2 as well...but maybe better to wait a little on that until I get the work done ;)
14:47 PacoLinux_ joined
jnthn "Look, you can natively call into C libraires!" could be nice marketing though :) 14:47
daxim libffi? 14:48
jnthn no, I'm not using libffi
I can't build the darn thing on my platform.
daxim bummer
colomon I'm hoping to do one on the mandelbrot script with a Gtk user interface -- assuming I can find time to write the code!
jnthn daxim: dyncall.org/ works out better on the building angle and is clean and well documented. And tested plenty "in the wild" too. 14:49
Plus has callback support...which is also maintiained on many platforms. Including Windows. 14:50
14:51 PacoLinux_ left
jnthn And the license lets us bundle it. 14:51
daxim :thumb: 14:54
[Coke] bundle++ 14:59
jnthn afk, beer shopping ;) 15:00
masak who is Father Chrysostomos, and how did he manage to stumble across a perl6 queue ticket and run it on Perl 5.10.1? :) 15:04
(not that I'm judging. I do more confused things than that with frightening regularity.) 15:05
15:07 lutok joined
daxim url? :) # curious 15:08
the father is PAUSE id sprout, also a frequent p5p-er 15:09
[Coke] masak: he's a core perl 5 contributor.
I would have assumed he was giving us the p5 view if he didn't carp about the signature being bead. 15:10
*bad
15:12 lutok left
masak well, it *is* bad, from a Perl 5 perspective. 15:12
I think data is insufficient whether he's aware he's in the p6 queue or not.
daxim: rt.perl.org/rt3/Public/Bug/Display...l?id=77170 15:13
[Coke]: oh, core perl 5 contributor! that's why his name sounded familiar!
15:16 Trashlord joined 15:26 rafl joined, bluescreen10 left 15:29 woosley joined 15:35 wtw left 15:41 bluescreen10 joined 15:47 Trashlord left, fsergot left 15:50 woosley left
[Coke] masak: I'm assuming he did a search for all recursion tickets and didn't restrict by queue or something. 15:50
rafl seen tadzik 15:52
aloha tadzik was last seen in #parrot 4 hours 22 mins ago saying "what particulary are you interested in?".
masak [Coke]: yeah, could be. 15:53
I think Perl 5 and Perl 6 manage to co-exist rather well these days. 15:56
Perl 5 is the "low-risk, low-gain" effort, and Perl 6 is the "high-risk, high-gain" effort. a bit like Google's division into "improve ECMAScript"/"develop Dart". 15:57
16:00 kaleem joined
[Coke] blogs.perl.org/users/alberto_simoes...rants.html -- normal (NON hague) grant deadline extended. 16:08
jnthn oh, my hague grant... 16:09
:)
[Coke] also: news.perlfoundation.org/2011/11/hag...lperl.html (dukeleto's pl/perl hague grant app) 16:11
masak interesting. 16:13
felher +1 on a "Perl 6 Advent Calendar 2011" :) 16:17
masak decommutes 16:21
16:21 mkramer joined 16:22 mkramer left, shanny joined 16:23 Guest2721 joined, mkramer joined 16:24 mkramer left 16:25 mkramer joined 16:27 shanny left
tadzik hello #perl6 16:30
16:30 tty234 joined
felher o/ tadzik 16:30
16:33 buubot_backup joined, tty234 left 16:34 kaleem left, tty234 joined 16:35 mkramer left 16:37 mkramer joined 16:41 birdwindupbird left 16:43 dukeleto joined, dukeleto left, dukeleto joined, dukeleto left 16:44 dukeleto joined, dukeleto left, dukeleto joined, dukeleto left 16:45 dukeleto joined, dukeleto left, dukeleto joined, dukeleto left 16:46 dukeleto joined, dukeleto left, dukeleto joined, dukeleto left 16:47 daniel-s left, dukeleto joined, dukeleto left 16:48 dukeleto joined, dukeleto left 16:49 dukeleto joined, dukeleto left, MayDaniel joined 16:50 dukeleto joined, dukeleto left, dukeleto joined, dukeleto left 16:51 dukeleto joined, dukeleto left 16:52 dukeleto joined, dukeleto left, dukeleto joined, dukeleto left 16:53 dukeleto joined, dukeleto left
mikemol Regarding a P6 advent calendar, I can, of course, do another entry on P6 resources and Rosetta Code. 16:53
16:56 kaleem joined 16:59 wolfman2000 left 17:14 uniejo joined 17:18 uniejo left 17:28 am0c left 17:29 am0c joined 17:32 kaleem left, orafu left, orafu joined, kaleem joined 17:35 PZt joined 17:38 thou joined 17:41 mkramer left 17:43 mkramer joined
tadzik djanatyn: ping 17:48
djanatyn tadzik: heyo 17:49
tadzik djanatyn: you were looking for Perl 6 tasks on GCI?
djanatyn Yeah! 17:50
I found a couple that I was considering working on, but they were all perl 5
so I was going to port Dancer plugins to Dancer 2. But if there's anything involving perl 6, I'd love to do that instead
tadzik djanatyn: we can figure out some nice tasks, I can then add and mentor them 17:51
PerlJam djanatyn: port Dancer 2 to Perl 6? :)
tadzik yes, that's nice. 48 hours enough? :)
PerlJam sukria already did all of the hard design work, porting to Perl 6 should be trivial! ;) 17:52
tadzik also, bailador already works :)
well, worked on b
17:53 cosimo_ joined 17:59 mj41 left
tadzik djanatyn: how would you feel about porting a module from Perl 5/ 18:02
?
something doable, Text::Something, or even Acme::Something :) 18:05
I wonder if Plack-related stuff would be approachable, since those are just subrefs returning an array
PerlJam Anything in the Plack space would be nice for web apps on Perl 6. 18:06
tadzik yeah
I wonder if we could easily write and use Plack::Test or such
I assume those are things which utilize the 'just a subref' patter, so it doesn't require any rocket science 18:07
[Coke] Tasks cannot be added to the TPF project until December mumble. 18:09
it is unlike last year when we could add them at will.
there is one more "task add" point at midway; no p6 tasks until then.
PerlJam [Coke]: er, what?
[Coke] so, checkout p5 or parrot in the meantime.
PerlJam: we (P6 folks) dropped the ball this year. 18:10
tadzik ok, that's fine 18:12
we just have to figure out tasks until Dec 16, and they'll get published then
[Coke] tadzik++ #optimist 18:14
tadzik :) 18:15
[Coke] is double checking his facts on #parrot
18:15 MayDaniel left
[Coke] parrot suggests that the rules may be slushy. If someone has a p6 task, they should consider trying to get it in (see #gci on perl.org) 18:16
18:20 PacoLinux_ joined 18:21 molaf left 18:23 mkramer left 18:24 mkramer joined 18:25 mkramer left 18:27 mkramer joined 18:29 envi_ left
[Coke] nope. not slushy. deadline has passed, per rafl on #gci 18:31
rafl if you guys can figure out a way of subverting any of the tasks we have right now into a perl 6 project, we'll be happy to do that as well 18:32
actually 18:33
we just got mail about a deadline extension
Nov 28. 8 UTC 18:34
get cracking!
18:35 mkramer left 18:36 PacoLinux_ left, PacoLinux_ joined 18:37 mkramer joined, mkramer left 18:40 am0c left
[Coke] rafl++ 18:42
18:50 kaleem left
djanatyn sorry for disappearing >_< 18:51
I had calculus class
tadzik: I would love to port a module from perl5. I have no idea how to, though
and sure, I'll port Dancer 2. I've used Dancer, but not Dancer 2
[Coke] wonders if there is work for a Mojo 6. 18:52
djanatyn sure :D
tadzik [Coke]: they claim so :)
istr __sri demands perldoc
djanatyn: wanna write perldocs? :) 18:53
djanatyn tadzik: yeah, I can write docs
did you see my blog post yesterday? that's about the peak of my ability
I know the basics of perl and have written a few scripts 18:54
tadzik djanatyn: nope, link?
djanatyn: a 'perldoc perlintro' would be a good start, I suppose
djanatyn djanatyn.github.com/2011/11/19/pars...hdays.html 18:55
I have a couple perl scripts on my github page - a few using SDL, and a little static website compiler using Template::Toolkit 18:56
19:01 spine joined 19:14 PacoLinux_ left 19:26 MayDaniel joined
[Coke] Quiet when jnthn has $dayjob work! 19:28
masak he's not the only one who has $dayjob work... 19:30
19:31 birdwindupbird joined, birdwindupbird left 19:32 PacoLinux_ joined 19:33 jaldhar left
tadzik sorry for the paste 19:33
We realize there have been some difficulties entering and editing your
initial tasks for GCI into Melange by today so we are extending the
deadline until next Monday, November 28th at 8:00 UTC. So please feel free 19:34
to add more tasks or edit your tasks as you see fit over the next week.
so we can figure out some Perl 6 tasks and add them
19:36 PacoLinux_ left
djanatyn tadzik: :D 19:42
Do you guys know of any other pre-university students working on perl6 stuff?
masak lue. 19:48
19:59 lumi_ joined 20:02 alester joined, wooden joined 20:05 lumi_ left 20:06 lumi_ joined, zostay joined 20:20 Moukeddar joined 20:28 ksi joined 20:37 MayDaniel left 20:38 PacoLinux_ joined 20:41 mj41 joined 20:44 Sarten-X joined 20:48 pomysl left 20:49 Sarten-X left, wallberg joined 20:52 estrabd joined 21:01 Exodist joined 21:02 Sarten-X joined 21:04 PacoLinux_ left 21:07 daxim left 21:11 Sarten-X left 21:13 cooper left 21:16 bluescreen10 left
dalek p/cclass: 0f3407b | diakopter++ | src/QRegex/P6Regex/ (2 files):
changes to support backslash endpoints
 21:19
21:21 Sarten-X joined 21:25 cooper joined, cooper left, packetknife joined 21:27 lutok joined 21:32 Sarten-X left 21:38 Moukeddar left 21:42 Sarten-X joined 21:45 Guest2721 left, pomysl joined, pomysl left, pomysl joined 21:54 spine left 22:00 lutok left 22:07 PacoLinux_ joined 22:11 PacoLinux_ left 22:18 PacoLinux_ joined
[Coke] begins a six day weekend wherein he might find some time to hack on perl6! 22:25
tadzik \o/
PerlJam [Coke]: have fun! 22:26
[Coke] see folks tomorrow.
22:29 envi_ joined, Sarten-X left 22:30 ksi left 22:34 PacoLinux_ left 22:36 Sarten-X joined 22:38 vmspb joined 22:41 Sarten-X left 22:47 Sarten-X joined 22:53 japhb_ joined
japhb_ jnthn, ping 22:53
phenny, ask jnthn Do you have any data about how many distinct STABLEs exist (created in the bootstrap, setting, etc.) for even an empty Perl 6 program? My brane is tanking on github.com/jnthn/6model/blob/maste...erview.pod again, and it occurs to me I lack numbers to contextualize with. 22:59
phenny japhb_: I'll pass that on when jnthn is around.
23:00 vlixes joined
jnthn japhb_: pong 23:00
phenny jnthn: 22:59Z <japhb_> ask jnthn Do you have any data about how many distinct STABLEs exist (created in the bootstrap, setting, etc.) for even an empty Perl 6 program? My brane is tanking on github.com/jnthn/6model/blob/maste...erview.pod again, and it occurs to me I lack numbers to contextualize with.
jnthn japhb_: I don't have an exact number. You can get a really good approximation by equating stable with type. 23:01
japhb_: That is, each knowhow/class/role/subset/enum/grammar declaration is an s-table
japhb_ phenny, ask jnthn Also, is there any way in the current 6model to handle native types nearly bare? (e.g. Mozilla's JS implementations where IIRC int, string, and object types are packed into special encodings of NaN in a 64-bit word, in union with a native double) 23:02
phenny japhb_: I'll pass that on when jnthn is around.
sorear good * #perl6
japhb_ oops, sorry, forgot to remove the phenny, ask when you showed up and I was still thinking. ;-)
jnthn japhb_: You could write a representation that does that kind of union-y thing if you wishes. 23:03
phenny jnthn: 23:02Z <japhb_> ask jnthn Also, is there any way in the current 6model to handle native types nearly bare? (e.g. Mozilla's JS implementations where IIRC int, string, and object types are packed into special encodings of NaN in a 64-bit word, in union with a native double)
jnthn *wished
japhb_: 6model does handle native types in various ways already
japhb_ OK, so it's not completely dependent on the first word being the STABLE pointer, it's just the way all existing repr's work?
jnthn japhb_: The P6opaque REPR knows how to store them "flat" inside another object's body, for example.
japhb_: er, that bit is important. :) 23:04
japhb_: For boxed objects
japhb_: However, if you statically know what type you have then you're free to keep things around on the stack, or in a packed array.
japhb_ Oh, OK, so you can't do less than two machine words for an object
jnthn japhb_: That's what REPR API 2 was all about.
japhb_: Not if you want it to be a standalone GC-able entity rather than in a memory space managed by something else, no. 23:05
japhb_ OK
23:05 Chillance joined
jnthn japhb_: Many REPR functions take pointers into the start of the object body though 23:05
japhb_: And operate independently of if the header is there or not 23:06
So long as you know the s-table *somehow*, you're all good
japhb_ Oh, excellent.
jnthn So you could do things like C#'s stackalloc.
You'd just have the memory space per call frame 23:07
And then have one static thingy that holds the s-table pointer
japhb_ One could in theory have a VM that used a single word packing like with Mozilla JS, and just used internally-known STABLEs for the native types, and only used "real" STABLEs for non-native types.
23:07 Chillance left
jnthn Also, if you statically know a REPR, you can inline all of the stuff it does too. They're non-virtual. 23:07
japhb_ ooh
nice
23:07 pomysl left
jnthn Yeah. I need to write a VM that does all of this some day. 23:07
japhb_ heh 23:08
jnthn In the back of my mind when I designed 6model was "how can I have all the dynamism, but still let a language like Java be able to JIT field accesses like a JVM can"
6model is massively influenced by gradual typing. 23:09
japhb_ Yeah, I've heard you say that enough times that I believe it. ;-)
jnthn Some days, so do I ;)
jnthn does a little Perl 6 related stuff 23:10
23:10 amkrankruleuen left
japhb_ On a vaguely related performance note -- in your various valgrind/cachegrind runs, did you ever get a sense of how much the performance of Rakudo is bounded by off-CPU memory access? (In other words, how cache-evil it is?) 23:11
23:11 Chillance joined, Chillance left
jnthn japhb_: I've not had chance to do cachegrind runs; it doesn't run on my platform. And the VS profiler that I do use for C doesn't give me cache miss info, sadly. 23:12
japhb_ :-(
Ah well.
jnthn japhb_: The GC is potentially the most cache horrible.
japhb_: Though I hunted down its worst behavior there and the Parrot folks patched it a while ago
japhb_ That's good at least.
jnthn (It used to stomp all over the heap when doing "is this a pointer" analysis and was probably pathalogical for anybody with a set associative cache...) 23:13
Basically it followed a link list of memory blocks. Which were allocated to probably hit page boundaries. 23:14
japhb_ Ewww.
And IIRC very few CPUs had high (or full) associativity past about L1 or so. 23:15
s/had/have had/
jnthn Yeah, that's my impression too.
23:19 skangas left, amkrankruleuen joined 23:20 mj41 left
sorear o/ jnthn 23:25
jnthn hi sorear 23:26
23:32 Ingis joined 23:33 envi_ left 23:37 tokuhirom joined 23:39 [particle] joined
dalek p/cclass: 74ca920 | jnthn++ | src/QRegex/P6Regex/ (2 files):
Simplify and correct the backslashes with .. in charclass patch. Needs more testing.
 23:41
23:45 whiteknight joined, whiteknight is now known as Guest19937
dalek p: 0f3407b | diakopter++ | src/QRegex/P6Regex/ (2 files):
changes to support backslash endpoints
 23:52
p: 74ca920 | jnthn++ | src/QRegex/P6Regex/ (2 files):
Simplify and correct the backslashes with .. in charclass patch. Needs more testing.
p: d4f6e5c | jnthn++ | src/QRegex/P6Regex/ (2 files):
Merge branch 'cclass'
kudo/nom: 55e5294 | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION to get char class improvements.
 23:54
kudo/nom: 4a57bd7 | jnthn++ | docs/ChangeLog:
Some ChangeLog entries.
jnthn phenny: tell moritz diakopter++ did some work to support backlash endpoints in char classes; I've just twiddled it a bit and updated NQP_REVISION. Hope this helps your JSON::Tiny work. 23:55
phenny jnthn: I'll pass that on when moritz is around.

Powered by Cro and the Raku® Programming Language.

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

Thank you!