Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_logs/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by AlexDaniel on 12 June 2018.
Geth roast: jstuder-gh++ created pull request #436:
Rakudo Issue 1918: Test Slurpy Positional params with .assuming		 02:07
MasterDuke timotimo: if sometimes an @-sigiled something is an NQPArray (so nqp::elems is fine) and sometimes it's an Array (so .elems is fine), is there a single say that will work for either case? 03:25
timotimo hm, .elems doesn't work with NQPArray? 03:26
MasterDuke: was it you who figured out the profiler missing one GC run by having its index be -1? 03:27
MasterDuke well, i noticed something 03:28
i think you actually found what was going on 03:29
and lizmat may have been the first to notice. or at least she noticed something, and then i went and maybe found something else?
timotimo i'm asking because my newest blog post was kind of implying "i fixed that" even though someone else basically did most of the detective work 03:30
and i didn't just want to take credit like that
anyway, all i think you can do is nqp::islist if NQPArray doesn't offer .elems; if it does, i'm hopeful that spesh will inline it gladly 03:31
BBL
MasterDuke i think of it sort of like i catalogued the symptoms, you made the diagnosis and the fix 03:32
later...
samcv i've gotten native int* and uint* types working in repl without the previous error after a fair amount of work in MoraVM and nqp 04:05
it shows a newer error certain times but it's a rakudo issue now instead of being a MoarVM and NQP issue 04:06
oh actually so there's no more issues at all for signed native ints now \o/ 04:07
but with uint's it will say "No such method 'perl' for invocant of type 'Perl6::Compiler'
AlexDaniel bisectable6: old=2018.05 class ThisName::Is::MuchToo::LongFor::Error::Messages { }; sub foo(Int $a) { }; foo(ThisName::Is::MuchToo::LongFor::Error::Messages.new) 06:02
bisectable6 AlexDaniel, Bisecting by output (old=2018.05 new=2fa08d5) because on both starting points the exit code is 1
AlexDaniel, bisect log: gist.github.com/4676bf89ae6bbcaf15...1711a83f99
AlexDaniel, (2018-05-20) github.com/rakudo/rakudo/commit/a7...74431d3d0e
AlexDaniel timotimo: ↑ there's the right one :)
AlexDaniel releasable6: status 07:46
releasable6 AlexDaniel, Next release in ≈1 day and ≈11 hours. 5 blockers. 87 out of 87 commits logged
AlexDaniel, Details: gist.github.com/0cfa0dc34a45e6ae64...352c7eb983
AlexDaniel lizmat++ # editing the changelog \o/ 07:47
lizmat and at the right place as well :-)
AlexDaniel btw anyone can do it to make sure your commits are properly changelogged: github.com/rakudo/rakudo/wiki/ChangeLog-Draft
lizmat Files=1239, Tests=76361, 316 wallclock secs (15.68 usr 5.60 sys + 2199.98 cusr 215.75 csys = 2437.01 CPU) 07:48
Geth nqp: 6c450557d4 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/build/MOAR_REVISION
[MoarVM Bump] 4eda279b5 Fix three memory leaks […]

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...g4eda279b5		 08:02
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...g4eda279b5
rakudo: 64394152cf | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 4 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....g6c450557d 6c450557d [MoarVM Bump] 4eda279b5 Fix three memory leaks […] e1ff93d5a Stub new atomic ops on the JVM backend 11cf77d33 Map two new atomic ops for MoarVM backend 6a27f87d2 Bump MOAR_REVISION for new atomic ops		 08:03
¦ rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....g6c450557d
lizmat weekly: unicode.org/consortium/adopted-cha...tml#b1F98B 08:43
notable6 lizmat, Noted!
AlexDaniel it's old news :) 08:45
lizmat it was new to me :-) 08:46
anyways, going to be afk because of commute to SLC &
AlexDaniel Geth: ver github.com/rakudo/rakudo/commit/a5...f1ced792cd 09:18
Geth AlexDaniel, version bump brought in these changes: github.com/perl6/nqp/compare/2018....g45ab6e373
AlexDaniel that makes no sense
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaah 09:21
Geth rakudo/rescalar: e31f830d30 | (Jonathan Worthington)++ | 4 files
Make Array use the the Scalar descriptor model

Instead of setting a `$!whence`, it now creates a descriptor that does the bind into the array's storage.		 09:50
Geth roast/post-release-2018.06: 9dc848d559 | (Jeremy Studer)++ | S06-currying/positional.t
Test Slurpy Positional params with .assuming

Testing that .assuming a sub with slurpy param compiles without error.
  See [Github Issue #1918](github.com/rakudo/rakudo/issues/1918)		 09:57
Geth rakudo/rescalar: 82702e9861 | (Jonathan Worthington)++ | 2 files
Update Rakudo::Iterator to not use $!whence
 10:10
AlexDaniel ok canary is green: travis-ci.org/perl6/rakudo-pkg-can...tification 10:15
and my sakefile goes through without any errors 10:16
so I guess these blockers that we have there already are the only ones
but we'll see
jnthn AlexDaniel: I left comments on one of the blockers; your feelings on what to do there are welcome 10:18
AlexDaniel oh yeah I've seen that, got carried away with other things though 10:19
let's see…
Geth rakudo/rescalar: 761c0bbdee | (Jonathan Worthington)++ | 2 files
Switch Array vivification to not use $!whence
 10:28
AlexDaniel left a comment (R#1923) 10:35
synopsebot_ R#1923 [open]: github.com/rakudo/rakudo/issues/1923 [regression][⚠ blocker ⚠] Regression related to MONKEY-TYPING and spesh-plugins merge
Geth rakudo/rescalar: 6ebc4b8723 | (Jonathan Worthington)++ | 2 files
Switch Hash vivification to not use $!whence
 10:38
Geth rakudo/rescalar: ada94199e5 | (Jonathan Worthington)++ | 4 files
Move ContainerDescriptor classes into BOOTSTRAP

So that we'll be able to refer to the stubbed Perl 6 types and not have to pass them in as an attribute every single time, which is not good on a hot path.		 11:13
rakudo/rescalar: 0273e33594 | (Jonathan Worthington)++ | 2 files
Avoid passing Array/Hash to descriptor constructor
nine timotimo: the issue with nqp::list($var) ?? nqp::elems($var) !! $var.elems is that $var will be evaluated 3 times causing redeclaration errors. Also it's not sure this is actually faster than the "unoptimized" +@list. So either we add the appropriate workaround in in method install_package, or we just revert the optimization 12:01
brrt \o 12:12
Geth rakudo/rescalar: 1cdf23f7ae | (Jonathan Worthington)++ | 3 files
Eliminate $!whence use from Stash, untyped Hash
 12:48
Geth rakudo/rescalar: de5df0a345 | (Jonathan Worthington)++ | 2 files
Eliminate $!whence use for object hashes

Also some code simplification and avoid looking up an attribute many times.		 13:11
timotimo removing whence from Rakudo_Scalar brings its total size down from 56 to 48, which is a much nicer number just aesthetically 13:28
KingWhenceless :D 13:31
timotimo oh my 13:35
Geth rakudo/rescalar: ef4385a63b | (Jonathan Worthington)++ | 5 files
Eliminate $!whence use for shaped arrays
 13:36
lizmat waves from AMS 13:37
KingWhenceless o/ lizmat 13:38
Have a safe flight (well, when it's time :)) 13:39
lizmat thank you 13:40
Geth rakudo/rescalar: 2d2fd86524 | (Jonathan Worthington)++ | src/Perl6/Metamodel/BOOTSTRAP.nqp
Eliminate $!whence handling logic in assignment
 13:43
lizmat boarding& 13:46
Geth rakudo/rescalar: fbbdaeff0c | (Jonathan Worthington)++ | 4 files
Eliminate the now-unused $!whence from Scalar

Making every Scalar container a pointer smaller. Since Array and Hash slots are also Scalar containers, they also get this much smaller per element (so on a 64-bit platform, a thousand element object Array gets 8000 bytes smaller).		 13:50
KingWhenceless Hurrah! 13:51
timotimo i published my blog post, this time with added cat picture (the one most of you have already seen): wakelift.de/2018/06/15/no-major-breakthroughs/ 13:51
jnthn timotimo++ 13:55
Got some other things to take care of right now. Next week's rescalar work will be unbusting the JVM build a bit more after these changes, and then digging in to using the spesh plugin mechanism to optimize assignment 13:57
And then probably lots of tweaking to improve the code quality 13:58
As in, quality of code generated after optimization. Not quality of the code I've just been writing. :P
brrt timotimo++ #blogging 14:10
dogbert17 timotimo++ blogging and cat picture 16:02
timotimo: if you want code examples where --profile behaves badly then I can supply a few 16:03
timotimo hah. depends on what kind of bad behaviour it shows 16:15
dogbert17 mostly monstrously large profiles 18:39
[TuxCM] This time from ansó: 18:40
Rakudo version 2018.05-88-g64394152c - MoarVM version 2018.05-124-g4eda279b5
csv-ip5xs0.880 - 0.882
csv-ip5xs-207.219 - 7.311
csv-parser23.435 - 24.336
csv-test-xs-200.462 - 0.476
test8.620 - 8.792
test-t2.265 - 2.299
test-t --race0.931 - 0.975
test-t-2040.230 - 40.871
test-t-20 --race13.556 - 15.200
timotimo dogbert17: if any of them are monstrous even if you disable inlining via the spesh env var, those would be interesting 19:29
jdv79 who's going to YAPC:NA? 21:38
that's soon, right?

Powered by Cro and the Raku® Programming Language.

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

Thank you!