01:01 FROGGS_ joined 04:52 colomon joined 06:17 camelia joined 06:27 camelia joined 06:30 camelia joined 06:51 brrt joined
brrt \o 06:51
07:06 FROGGS_ joined 08:32 zakharyas joined 09:30 cognome joined 10:15 cognome joined 10:46 cognome joined 10:47 cognome joined 10:50 avuserow joined 10:53 leont joined
timotimo t.h8.lv/p6bench/2014-09-25-two_mont...akudo.html - t.h8.lv/p6bench/2014-09-25-two_mont...story.html 11:07
brrt: -^
nwc10 timotimo: is your setup good enough to also the setup/starup for perl 5 loading Moose? 11:10
In that, those are startup + runtime benchmarks, aren't they? 11:11
timotimo no, they are runtime - startup
nwc10 ah OK. cool. I'm being, um, LTA
thanks
timotimo it would be pretty cool if you could teach the compilers.pl about using Moose and the components.json about installing Moose (and Data::Alias and the other thing that's used in one of the microbenchmarks) 11:12
11:21 camelia joined 11:48 cognome joined 12:31 cognome joined 12:46 cognome joined
dalek arVM: 8b43ef6 | moritz++ | Configure.pl:
[Configure] document --asan option
 13:08
arVM: 8a5d684 | moritz++ | Configure.pl:
[Configure] synchronize documentation of command line options
 13:17
13:46 kjs_ joined, cognome joined 14:01 colomon joined 14:46 cognome joined 15:28 Ven joined
japhb nwc10, timotimo: The big thing I need for doing benchmarks with Data::Alias, Moose, etc. is a way to install them -- and particular versions of each -- with zero configuration, preferably with the ability to network fetch separately from build/install into the locally-built perl5's. If Moose and all its prerequisites are all available via git with proper tagging, then we can probably treat them as components and 15:33
we can just have the module installs be part of the build procedure for perl5. 15:34
Someone would need to track that down, and figure out how the prerequisites for Moose have changed over time, and if they need to be changed based on perl5 version, or if the same list of modules can be installed for every supported perl5 release (which has so far been 5.10+, but we can change if really necessary). 15:36
15:46 cognome joined 15:50 FROGGS[mobile] joined 16:09 colomon joined 16:46 cognome joined 16:47 FROGGS joined 17:06 rurban joined
rurban moritz: typo in your —asan commit 17:06
"suport"
moritz rurban: thanks 17:07
dalek arVM: 078176d | moritz++ | Configure.pl:
fix typo, rurban++
17:46 cognome joined 17:58 kjs_ joined 18:04 cognome joined 18:17 kjs_ joined 19:02 colomon joined 19:08 Ven joined
nwc10 japhb: I don't have any real insight to help you with that 19:42
timotimo brrt: t.h8.lv/p6bench/2014-09-25-many_rak...story.html - t.h8.lv/p6bench/2014-09-25-many_rakudos.html
and this one is without perl5: t.h8.lv/p6bench/2014-09-25-many_rak...tory2.html 19:45
moritz does anybody know what made while_push_join so slow? 19:47
or while_push, for that matter
FROGGS moritz: yes 19:50
there was a case of a single item push that was really slow
and TimToady hunted it down to a push candidate and removed it 19:51
which in turn made out benchmarked case slow
jnthn afterwards found a way to satisfy both cases, and made it slower than the original version :o) 19:52
moritz uhm 19:55
that doesn't sound like a happy ending :(
FROGGS err 19:59
s/slower/faster/
*g*
hehe
timotimo since i'm currently not sitting at that computer, i can just run HEAD on top of that 20:01
FROGGS moritz: this was the patch that made both microbenchmarks *faster* in the end: github.com/rakudo/rakudo/commit/abc97dbcc2 20:04
moritz FROGGS: thanks 20:05
FROGGS six days old...
moritz are the various MoarVM env variables documented somewhere?
FROGGS I don't think so
nwc10 I should add that to my list of "mistakes not to repeat from Perl 5" 20:06
ie, please, someone, document them
FROGGS last time I grepped for DISABLE I just found it in /src
nwc10 and hence start to see how long the list is getting
moritz where should I document them? 20:10
20:11 brrt joined
brrt timotimo++ :-) 20:11
nwc10 moritz: good question, to which I don't know the answer. Perl 5 has documentation intended to become man pages. I don't know if MoarVM has a man page yet 20:12
FROGGS moritz: doc/env_variables.md?
moritz FROGGS: in the long run, a man page might be better indeed
but it seems that md2man is only available as a ruby tool 20:13
so I'd write it in POD, and use pod2man
if nobody objects
brrt also, is moar-jit slower in most of these benchmarks? 20:14
i don't object, i just think md is often easier than pod, and also i don't know pod that well 20:15
PerlJam moritz: you could always write a perly version of md2man first ;> 20:16
moritz PerlJam: I'm not into deep yakshave diving mode
brrt and while_concat is missing for moar-jit?
FROGGS m: sub foo($a) { }; foo() 20:19
camelia rakudo-moar 7469aa: OUTPUT«===SORRY!=== Error while compiling /tmp/JMYLrHsAX4␤Calling 'foo' requires arguments (if you meant to operate on $_, please use .foo or use an explicit invocant or argument)␤ Expected: :(Any $a)␤at /tmp/JMYLrHsAX4:1␤------> [32…»
brrt in fact on those benchmark moar-jit seems to do very bad 20:29
dalek arVM: fc4151a | moritz++ | docs/moar.pod:
Add beginnings of a manual page in docs/oar.pod

I use pod here instead of markdown, because there is a widely available pod2man tool.		 20:30
PerlJam moritz: you've at least one typo "contributrs". I'd fix it myself, but it seems like a lot of work to make a PR for just that little change :) 20:34
dalek arVM: 552e5aa | moritz++ | docs/moar.pod:
fix typo, PerlJam++
 20:35
20:45 rurban left 20:55 kjs_ joined 20:59 brrt left 21:22 flussence joined 22:34 cognome 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!