Xliff timotimo: Yep that. 03:16
lizmat Files=1336, Tests=113586, 221 wallclock secs (29.40 usr 8.43 sys + 3102.36 cusr 290.46 csys = 3430.65 CPU) 09:20
sena_kun 09:43
Geth ¦ problem-solving: JJ assigned to jnthn Issue Ecosystem name resolution (or squatting) problem github.com/Raku/problem-solving/issues/229 09:44
rakudo/precomp_store_parallelism: 4 commits pushed by (Stefan Seifert)++ 10:07
timotimo Xliff: yeah, that's an accurate assessment 10:44
Xliff timotimo: ah. Thanks. 10:47
OK. Maybe bug?
github.com/rakudo/rakudo/blob/mast...s.pm6#L512
Why $thunk() -- This results in No such method 'CALL-ME' for invocant of type 'Str' 10:48
m: my $a = 'a'; $a()
camelia No such method 'CALL-ME' for invocant of type 'Str'
in block <unit> at <tmp> line 1
Geth ¦ problem-solving: holli-holzer assigned to jnthn Issue rx:s/ / is considered a null regex github.com/Raku/problem-solving/issues/230 10:55
Xliff Method 'content' must be implemented by CompUnit::Repository::Distribution because it is required by roles: Distribution. 10:58
^^ From fresh build attempt
So it looks like the parens around $thunk ARE needed. 11:01
Ah, so it's trait_mod:<handles>($attr, sub { <expr }) 11:08
However that still isn't working at run-time.
ENOTABUG 11:09
Geth rakudo/master: 5 commits pushed by (Stefan Seifert)++, niner++ 11:23
Altai-man nine++ 11:26
Xliff Altai-man++ 11:36
timotimo Xliff++ 11:37
nine timotimo++ 11:38
lizmat nine: does that imply that tools/build/install-core-dist.raku could be made to install all modules in parallel ? 11:47
nine lizmat: actually, yes
lizmat looks like that needs a change in CU::Staging, right ? 11:48
nine lizmat: from a quick look, this actually means that you'd have to make dist installation parallelized in general, which would be quite awesome 11:49
github.com/rakudo/rakudo/blob/mast...n.pm6#L270 11:50
lizmat do the $precomp.compile in a start block and wait for all the promised ? 11:51
*promises
nine something like that, yes 11:52
[TuxCM] Rakudo version 2020.08.2-59-ge65466fcd - MoarVM version 2020.08-91-g590bac47e
csv-ip5xs0.954 - 0.989
csv-ip5xs-209.803 - 10.041
csv-parser25.481 - 30.616
csv-test-xs-200.397 - 0.399
test7.561 - 7.856
test-t1.868 - 1.906
test-t --race0.830 - 0.851
test-t-2031.670 - 32.964
test-t-20 --race9.263 - 9.447
 11:59
Altai-man oooh, this sounds juicy 12:04
releasable6, status 12:09
releasable6 Altai-man, Next release in ≈13 days and ≈6 hours. 1 blocker. Changelog for this release was not started yet
Altai-man, Details: gist.github.com/132920cfc4d7244fb0...5b1dd15311
lizmat nine: could you comment on this comment? 12:13
# the following 3 `for` loops should be a single loop, but has been
# left this way due to impeding precomp changes
nine no 12:14
lizmat ok
nine I'd ignore it
lizmat any reason why it does "%provides.sort" why the sort ? 12:15
timotimo impeding or impending? 12:16
nine Looks like a reproducible builds fix 12:17
timotimo that sounds like a good assumption for every appearance of "sort" 12:20
nine It's also what my commit message indicates. Though I honestly cannot say how the order would change the result and I didn't mention that :/ 12:21
timotimo insort it and see what happens ;) ;) 12:22
unsort*
lizmat afk for a few hours& 12:42
Geth rakudo/parallel-module-install: 41e7f5e3a2 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Step #1: streamline CURI.install

  - use binding where possible, ensures no changes and save allocations
  - fix indentation on protect block / localization dynvar blocks
  - remove unnecessary intermediate variables
  - rename $lock to $repo-lock, to prevent confusion with $!lock
  - remove unnecessary "is copy" in signatures
  - remove unneccssary .IO on already IO::Path objects
  - only sort %provides once, work using keys from there on
  - replace .values[0] with .values.head to prevent unnecessary caching
  - some code re-ordering to prevent unnecessary work if next fires		 20:28
rakudo/parallel-module-install: 4eb221dda5 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Step #2: naive parallelization of module installation

Basically run the precomp in a start block, collect the promises and await them. Results for installation of core modules is meagre. Installing something like App::Mi6 goes from 18 -> 15 seconds wallclock. Which is... disappointing. I wonder why this doesn't improve more.		 22:45
timotimo lizmat: can you post like a module debug log or something of mi6 or core modules? 22:55
there must be something somewhere that forces serialization? 22:56
MasterDuke Xliff: does the ^^^ branch do anything for your modules? 23:11

Powered by Cro and the Raku® Programming Language.

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

Thank you!