01:20 buggable joined 01:23 buggable joined 01:26 buggable joined 01:48 ilbot3 joined 03:08 geekosaur joined 03:42 mtj_ joined 05:45 geekosaur joined 06:46 geekosaur joined 06:53 domidumont joined 07:08 domidumont joined 07:10 domidumont joined 07:13 domidumont joined
timotimo i notice moarvm.org's measurements folder hasn't updated since march 09:22
timotimo AFK
09:47 robertle joined 09:56 robertle_ joined
Geth MoarVM: a0992f3872 | (Jonathan Worthington)++ | 5 files
Fix newline translation in various encodings.

They didn't go through the normalizer due being automatically in NFG. This meant they missed out on \r\n -> \n translation.
11:41
nwc10 it's next week already? gosh! 11:43
jnthn That wasn't the thing I said I'd put off until next week :) 11:44
13:32 zakharyas joined
Geth MoarVM: 9518234de8 | (Jonathan Worthington)++ | docs/ChangeLog
Add 2017.06 ChangeLog entries.
14:09
MoarVM: e96226e4ca | (Jonathan Worthington)++ | VERSION
Bump VERSION.
14:13
jnthn Checking of ChangeLog welcome 14:14
Will cut release in an hour or so :) 14:16
15:23 domidumont joined 15:43 btyler joined
jnthn "or so" :) 16:07
16:14 nebuchadnezzar joined
jnthn www.moarvm.org/releases/MoarVM-2017.06.tar.gz 16:24
eveo: ^^
eveo jnthn++ thanks
jnthn Pretty nice changelog this month :) 16:25
timotimo i read on reddit that some changes that would make I/O faster were going to be kept until the next release? 17:30
jnthn wonders what changes those are :P 17:41
I mean, it's true in so far as more improvements are possible 17:42
But it's not like we've anything sat in unmerged branches
There was no intentional "we'll hold this back just in case", though :) 17:43
jnthn found www.tomshardware.com/news/intel-mes...34806.html an interesting read 17:52
17:58 d4l3k_ joined
timotimo i can't find the reddit thread any more where i saw that claim 18:17
Geth moarvm.org: d262b48e64 | jnthn++ | 2 files
Update size for 2017.06 release.
18:26
jnthn bah 18:27
*site :P
I hope there's less typos in the updates
timotimo can you figure out why the benches ain't updating any more? 18:31
18:52 dogbert17_ joined
lizmat timotimo: perhaps it was me on the mailing list 19:30
timotimo ah, yes! it was on the mailing list! 20:00
lizmat which I wrote while being under the impression of what I wrote :-) 20:06
jnthn timotimo: Hm, think I saw what's wrong and fixed it. Guess we find out tonight :)
21:40 mj41 joined
mj41 jnthn: Ahoj. I just noticed this regression gist.github.com/mj41/20c7056656428...2a2593L135 21:44
gist.github.com/mj41/20c7056656428...282a2593L4
also bad with disabled JIT gist.github.com/mj41/20c7056656428...2a2593L107 21:46
But you probably know :-). So happy hacking. I will retest after work is done. 21:47
Dobrou noc.
jnthn mj41: Diky moc :)
mj41: I'm aware there's a regression on output; I've got a plan for solving it, but didn't get to it yet :)
Thanks for supplying data on the magnitude though.
It's a bit worse than I'd thought
mj41 In this case it make sense gist.github.com/mj41/6761c021135ae...1-txt-L138 21:57
It's weird from my amateur point of view that it is somehow related to the loop without output inside. See gist.github.com/mj41/20c7056656428...1-txt-L223 21:58
But my part is done. Will backlog tomorrow :-). 22:01
Díky moc za všechnu tu práci. Dobrou. 22:04
jnthn :-) 22:05
Dobrou noc o/
timotimo that loop mj whipped up is being treated the worst way we can 23:03
for example, in order to do the comparison, we box the int and then compare. then we create a p6bool from the result which we then immediately unbox_i again to check the result value 23:04
there's also an attempt being made to sink something 23:05
the $i is passed as an intlexref to the + operator 23:07
there's 5.9k GC runs 23:08
all of them about 0.9 miliseconds long, as they don't promote a single byte
timotimo goes to bed 23:10
23:21 AlexDaniel joined