github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
timotimo hah, i wanted to show in the blog post how you can get around "method call on native str boxes an object for you" by using, for example, the index sub instead of the index method on Str 00:17
turns out the implementation of the index sub is currently:
multi sub index(Cool $s, Cool $needle, Cool $pos) { $s.index($needle,$pos) } 00:18
timotimo join not returning 8bit strings when it could is currently letting this program consume almost exactly 1 gig when i think it could be a quarter of that 01:45
timotimo i would like a bit of blog post review please :3 wakelift.de/p/31a96e34-887c-468e-8...4d448ec1f/ 02:03
3.4k words 02:04
MasterDuke timotimo: 'say "Spliting file ...";' doesn't really matter, but should be "Splitting" 02:22
geekosaur "spiting file" :p
MasterDuke timotimo: "there's also sub forms" should be "there are also sub forms" 02:27
MasterDuke very nice post, timotimo++ 02:31
AlexDaniel` squashable6: next 04:53
squashable6 AlexDaniel`, ⚠🍕 Next SQUASHathon in ≈5 hours (2018-09-01 UTC-12⌁UTC+14). See github.com/rakudo/rakudo/wiki/Mont...Squash-Day
nwc10 oooh, it's almost time for 05:10
/win 1
buggable nwc10, Thank you for entering Accidental /win Lottery! The next draw will happen in 18 hours, 49 minutes, and 33 seconds
brrt ohai 06:25
brrt timotimo++ nice post 07:28
nine I just realized that if a NativeCall user often gets strings from the native library just to pass them back to some native code, it simply shouldn't declare the return and arg types as Str but as Pointer. 07:34
Whether this is a good strategy for Inline::Perl5 or if it would just optimize for test-ip5xs at the cost to more realistic use cases will be hard to find out unfortunately :/ 07:36
brrt hmm 07:38
we assume test-ip5xs is a realistic post?
eh test
nine Well it's realistic, but also a micro benchmark. 07:40
jnthn samcv: Please just take them as is and do the release. I think if the changelog is causing this much work/delay we shouldn't bother with it. 09:31
samcv ok, got it
next releaese i'm going to change the way i do the changelog so hopefully this won't happen again 09:32
jnthn I'm honestly not sure who reads them
samcv well i read changelogs :)
jnthn Ah :)
samcv anyway i must go to bed. have a phone call in the morning about a job
jnthn Rest well, and good luck with the call :)
samcv AlexDaniel`: i should release tomorrow then
night o/ 09:33
brrt good night and good luck 09:38
brrt is nearly done with the expr compiler refactor
operand type checking turns out a major pain, because of variadic
brrt grumbles because it turns out that the whole set of 'atkey' templates is wrong 09:48
... which is why it is good that we have stricter validity checks now
I might even add carg validity checking at some point 09:49
jnthn Better we find them that way than with crashes and similar 09:50
brrt so the curious thing is... we didn't find them with crashes :-)
nine The changelog also ends up in the openSUSE RPM packages. But there's also a limit to the number of allowed lines, so a curated changelog makes my life much easier than an unfiltered commit list. 09:53
brrt nods 09:54
who volunteers? ;-)
blegh, core dump 10:00
nine Fun fact: csv-ip5xs.pl does not in fact pass strings it got from the native code back to native code. It reads lines from STDIN and passes those. No idea why I've always thought this different. It's not exactly super obscure code :) 10:04
timotimo what kind of wrong did the atkey templates have? :o 10:23
brrt they had a $0 for the object 10:48
while they needed $1 10:49
timotimo also, i hope you don't just say "nice post" because it's so many words :D
nine doesn't think "nice post" is many words 10:50
timotimo www.youtube.com/watch?v=uR7euFrhMAQ
Geth MoarVM/template-compiler-refactor: e7ee6b4b9f | (Bart Wiegmans)++ | 5 files
[JIT] New template compiler

  - Single-pass compilation (no separate 'validate_template' step
  - Better type and format checking (operands should be expressions or
   references, parameters should be barewords, numbers or macros)
  - Write-out parameters need to be refered with a \$ sigil, like \$0,
   and templates without output operands should not get a '!' suffix		 13:21
MoarVM: bdw++ created pull request #950:
[JIT] New template compiler		 13:23
timotimo wakelift.de/2018/08/31/faster-fasta-please/ - i published my blog post! 16:52

Powered by Cro and the Raku® Programming Language.

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

Thank you!