Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2015-01-28

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:48 ilbot3 joined #moarvm
02:48 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
03:52 vendethiel joined #moarvm
04:48 sivoais joined #moarvm
07:10 vendethiel joined #moarvm
07:39 kjs_ joined #moarvm
08:11 FROGGS joined #moarvm
08:15 vendethiel joined #moarvm
08:57 zakharyas joined #moarvm
08:57 brrt joined #moarvm
09:29 kjs_ joined #moarvm
09:35 vendethiel joined #moarvm
09:44 kjs_ joined #moarvm
10:44 vendethiel joined #moarvm
11:05 brrt joined #moarvm
11:07 rurban_ joined #moarvm
11:08 vendethiel joined #moarvm
12:02 kjs_ joined #moarvm
12:21 brrt joined #moarvm
12:30 jnthn Maybe p6invokeunder is something we should not inline...
13:03 btyler joined #moarvm
13:43 brrt joined #moarvm
14:09 brrt jnthn - fwiw, the issue seems to be that the spesh candidate is supposed to have a two-args-callsite, and the second of these args is really an integer
14:10 brrt now that i'm talking about it, it seems plausible that that second argument may not be an integer at all
14:10 brrt just some random value
14:10 brrt and that the args buffer is really smaller than this
18:40 rurban_ joined #moarvm
19:03 brrt joined #moarvm
19:03 FROGGS joined #moarvm
19:30 muraiki joined #moarvm
19:30 muraiki hi all. I'm getting an error trying to build moar on (an admittedly old) freebsd using rakudobrew
19:31 muraiki oh wait, I see
19:31 muraiki it wants clang, and I only have gcc
19:32 muraiki probing whether your compiler thinks that it is gcc  clang: not found
19:32 muraiki Can't compile simple gcc probe, so something is badly wrong at build/probe.pm line 92.
19:33 nwc10 muraiki: try running Configure.pl as
19:33 nwc10 Configure.pl --cc=gcc -ld=gcc
19:33 nwc10 (plus any other options you had)
19:33 jnthn I think --compiler=gcc is enough to get it to set the cc, ld, etc.
19:34 nwc10 ah OK.
19:34 nwc10 I was searching back through command line history on a FreeBSD box
19:34 nwc10 anyway, unambiguous conclusion is "it should work, but you'll have to tell it not to default to clang"
19:36 nwc10 sigh, build system. We ought to fix it so that ... and then I think "step back, maybe there's a bigger set of things to improve"
19:36 nwc10 no, more like "step back, maybe it's more effective in the long term fixing the bigger picture problems"
19:36 muraiki thanks, I'll try that
19:36 nwc10 anyway, sorry that it doesn't just work right first time with the defaults
19:36 muraiki it's ok. I'm not a freebsd kind of person, but it's what I have at work
19:36 muraiki and this is an ancient version, too
19:36 nwc10 oooh that's an interesting work
19:37 nwc10 most places I've worked that go open source OS are some or other linux
19:37 nwc10 even the place where the CTO likes FreeBSD on his personal machines
19:37 muraiki yeah, this place has a long history of freebsd usage, although perhaps now 10-20% are linux
19:38 nwc10 the gcc that certainly did built OK at one point is "gcc (GCC) 4.2.1 20070719  [FreeBSD]"
19:39 muraiki nice, I have the same oen :)
19:50 muraiki ok, I got "Configuration SUCCESS." but then I also got two errors: ""Makefile", line 30: Missing dependency operator "Makefile", line 35: Need an operator make: fatal errors encountered -- cannot continue"
19:50 muraiki oh, I wonder if I have to use gmake
20:26 muraiki ok, it worked with --compiler=gcc --make=gmake --optimize
21:07 avarab joined #moarvm
21:14 timotimo moarvm likes --optimize=3
21:15 timotimo i think --optimize is the same as --optimize=1 and no --optimize is the same as --optimize=2 nowadays
21:15 jnthn I bumped the default to 2
21:16 jnthn Let's give that some time to make sure there's no fallout, then do 3.
21:16 jnthn Otherwise if there is fallout we won't know which is to blame.
21:17 timotimo i have been compiling and running and spectesting with -O3 for a long time :)
21:17 timotimo i saw we are waaaaayyy slower than rakudo-jvm and rakudo-parrot on one of the benchmarks
21:17 timotimo deep_scan_for_string_literal or something
21:17 timotimo 400x slower or 500x
21:18 timotimo (parrot is only 3x or 4x slower than jvm here)
21:18 timotimo haven't had the opportunity to look more closely
21:18 jnthn Wow
21:18 jnthn That's gotta be some algorithmic mis-fortune...
21:19 jnthn deep_scan_for_interpolated_string_var ?
21:19 timotimo i think so, yes
21:26 timotimo i'm getting impatient WRT perl6 performance :\
21:26 timotimo but i suppose most of it is still GLR-related
21:31 muraiki what's the timeline for GLR?
21:40 FROGGS muraiki: '2015'
21:40 muraiki that's the timeline for everything! :)
21:40 FROGGS muraiki: but pmichaud++ said that he is going to work on it mid February, so, we'll see
21:41 muraiki ok, I'm still learning about this stuff -- I was just curious
21:44 timotimo mid february? i like that
21:45 timotimo now that i have an actual time frame, i should be able to keep my impatience under control
21:49 FROGGS :o)
21:51 japhb FROGGS: But he keeps bumping the "I should be able to work on it in this window" date.  Encouragingly, he said this time that if he misses the window, someone else can pick up the task.  So hopefully he will be able to, but if not, Rakudo isn't blocked.
21:53 FROGGS japhb: true
21:53 FROGGS but who wants to pick up *that* task?
21:53 FROGGS I mean, jnthn++ can't do all three big TLA :S
21:53 lizmat I guess if we must, then we must
21:54 lizmat I've been digging in all of the xxIter stuff lately, to familiarize myself with that
21:54 lizmat just in case
21:55 timotimo brave
21:56 lizmat perhaps, but I would hope not to have to do that by myself
21:56 timotimo agreed, it's a big chunk of work
22:13 woolfy1 joined #moarvm
22:13 woolfy1 left #moarvm
22:15 woolfy1 joined #moarvm
22:40 jnthn I don't mind too much if it falls to me, but I'm doing native arrays and NFG first.
22:41 jnthn And I still think pmichaud++ would make a far better job of it.
22:42 jnthn I mostly just have the advantage of being a more stubborn debugger than most :P
22:44 FROGGS I also don't tend to give up early, but I clearly lack a CS study and a lot of knowledge
22:47 jnthn CS studies taught me how knowledge I don't have... :)
22:47 FROGGS the what? :D
22:48 jnthn You have to know some fields exist before you can know you don't know about them :)
22:48 FROGGS gotcha
23:41 vendethiel *g*

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary