02:54 andres_ joined 02:56 andres_ left 05:08 Kaiepi left 05:09 Kaiepi joined 07:31 sena_kun joined
nine '/home/nine/rakudo/install/bin/nqp-m' --module-path=blib --ll-exception --target=mbc --output=blib/Perl6/World.moarvm gen/moar/World.nqp 08:00
Missing or wrong version of dependency 'gen/moar/stage2/NQPHLL.nqp' (from 'gen/moar/Pod.nqp')
08:00 Kaiepi left
nine So...the build is broken again? 08:00
08:07 jmerelo joined
jmerelo Distribution in age groups of responders of the Raku survey through the years github.com/Raku/survey/blob/master...-years.png 08:08
Biggest difference seems to be in the 18 thru 40 age group, which has lost ~20% In fact, that pretty much explains the difference in total number of answers 08:10
08:17 domidumont joined
jmerelo And prior use of Perl github.com/Raku/survey/blob/master...-years.png Those tho have used Perl have decreased more than those who haven't, but it's still the bigger majority 08:20
sena_kun nine, I wonder if nqp wants the same treatment as github.com/rakudo/rakudo/commit/a5...eff7810242 08:21
08:24 domidumont left
nine m: package Foo { package Our::Bar {}; package Their { package Baz {} } }; say Foo::Our.^name; say Foo::Their.^name 08:57
camelia Our
Foo::Their
nine I dare say this difference in names (one without and one with the containing package as a prefix) is a bug. And one that causes some of the remaining issues with in-process precompilation
lizmat Files=1336, Tests=113586, 241 wallclock secs (30.78 usr 9.36 sys + 3328.17 cusr 319.26 csys = 3687.57 CPU) 09:55
nine Well the fix looks simple and passes tests and spectests 10:06
Geth rakudo/fix_name_of_stub_packages_in_multi_part_declarations: dae8383a20 | (Stefan Seifert)++ | src/Perl6/World.nqp
Fix stub packages in multi part declarations only knowing part of their name

In package Foo { package Bar::Baz { } } the created stub package for Bar had only "Bar" as longname instead of the expected "Foo::Bar" like you'd get if the Bar package was created explicitly. Fix by using the target package's name as starting point when assembling the longname.
10:09
rakudo: niner++ created pull request #3926:
Fix stub packages in multi part declarations only knowing part of the…
10:12 jmerelo left 11:02 Altai-man joined 11:05 sena_kun left
Geth rakudo/rakuast: 301f2d1890 | (Elizabeth Mattijsen)++ | src/Raku/ast/expressions.rakumod
Remove unnecessary whitespace
11:43
rakudo/rakuast: fed18ca0a5 | (Elizabeth Mattijsen)++ | t/12-rakuast/var.t
Test deparsing most of variable declarations
nine Simple...but not actually entirely correct 11:44
lizmat ?
nine package GLOBAL::Foo { } should still be called Foo 11:45
Geth rakudo/fix_name_of_stub_packages_in_multi_part_declarations: 51db5341a7 | (Stefan Seifert)++ | src/Perl6/World.nqp
Fix stub packages in multi part declarations only knowing part of their name

In package Foo { package Bar::Baz { } } the created stub package for Bar had only "Bar" as longname instead of the expected "Foo::Bar" like you'd get if the Bar package was created explicitly. Fix by using the target package's name as starting point when assembling the longname.
11:48
nine This works better
lizmat good catch! 11:49
[Tux] This one is still from yesterday: 13:22
Rakudo version 2020.08.2-79-gb857aee9d - MoarVM version 2020.08-93-g1fdfdf59c
csv-ip5xs0.992 - 1.016
csv-ip5xs-209.892 - 10.580
csv-parser26.387 - 28.251
csv-test-xs-200.386 - 0.401
test7.919 - 8.567
test-t1.908 - 2.001
test-t --race0.848 - 0.874
test-t-2033.216 - 33.934
test-t-20 --race9.271 - 10.099
Rakudo version 2020.08.2-81-g8e6f7f0d2 - MoarVM version 2020.08-93-g1fdfdf59c 13:36
timotimo so how exactly does relocatable rakudo work? 13:55
right now i get "NQP_HOME is invalid"
hum. 13:57
setting rakudo and nqp home env vars is fine and all, but now it's trying to invoke a rakudo from a wrong path to do precomp or something?
ok, i got further this time, cool. 14:00
nine Down to 3 spectest and 2 test files failing with in-process precomp 14:59
MasterDuke is it noticeably faster?? 15:01
15:03 sena_kun joined
nine Core module installation in 8.1s vs. 11s on master 15:04
15:05 Altai-man left
MasterDuke nice. and it's profileable now, right? 15:06
nine think so 15:09
MasterDuke cool beans 15:10
timotimo yo MasterDuke you wanna test a random thing i made? 15:20
MasterDuke i probably can 15:21
nine Type parameterization is still an issue when multiple files are precompiled by the same process. A parameterization of a setting type (like Hash[Any:D,List:D]) created during one precompilation is also found when a later compilation needs that type. But it will already have been added to a different SC 15:24
timotimo so we need many parameterization caches 15:27
one for anything used in the compiler
and one for each compilee?
nine There's another factor: the cache is not just for performance. It's also there so that we do get the same type object for a parameterization, otherwise those types wouldn't match 15:29
timotimo yeah
nine I wonder though why it won't just be recorded as a dependency and instead I get STable Perl6::Metamodel::ClassHOW+\{<anon>}+\{<anon>} does not exist in serialization context
timotimo nine, got a minute to run a binary for me? 15:30
nine sure
timotimo u.setxkbmap.de/#xHBkV2CqBsD2rJHzZEgDDA
nine ===SORRY!=== Error while compiling /tmp/.mount_MoarPe5H1oFu/usr/rakuapp/service.p6 15:32
Failed to spawn process /usr/rakudo/bin/rakudo: no such file or directory (error code -2)
timotimo now why would it go ahead and do that 15:34
i fixed that by removing -I . from the commandline 15:35
nine To distract me from noticing your evil rootkit taking over my machine?
timotimo i wonder if it just found it in my global system instead
it's a rakuutkit
what does RAKUDO_MODULE_DEBUG look like? 15:36
nine gist.github.com/niner/48dfe68f3b9d...f844aae6f0
timotimo u.setxkbmap.de/#6-i34xea2w82N20l2ha5LQ 15:38
1 RMD: Precompiling /tmp/.mount_MoarPe5QqXwT/usr/bin/../rakudo/share/perl6/site/sources/6D2B9EAD4DC4FCDFF835D33B0CA4F37530A7DDCE into /home/nine/.raku/precomp/2B6602E5A8534CC907DE2D4F2BA95B45CA492917/6D/6D2B9EAD4DC4FCDFF835D33B0CA4F37530A7DDCE.bc ( ) 15:39
well that isn't good
it should(. do that, right?
it doesn't seem to load from my home directory on my machine 15:40
nine Probably because you don't have a conflicting cro version installed into your home repo 15:41
timotimo got a suggestion how i should proceed? 15:42
if you want, you can put --appimage-extract as argument and it'll unpack a folder called "squash-fs-root" or so, and you can ./squashfs-root/AppRun to run the thing
it'll go blah-root/usr/bin/moarperf-launcher-script or whatever i called it 15:43
nine if the goal is to provide AppImages of rakudo based apps, you got to get rid of that home repo in the repo chain 15:44
timotimo right, is there a way to do that without patching the rakudo i build?
though i can of course do that
nine I don't think there is one yet
timotimo oh, what does RAKUDO_PRECOMP_WITH do? 15:47
nine That's used to pass on the current repo chain to precomp processes 15:48
It's part of what I'm trying to get rid of :D
timotimo ah
looks like my patch successfully tosses out home dir precmomp 16:03
MasterDuke isn't that a pokemon? 16:04
16:07 MasterDuke left
timotimo nine: github.com/timo/moarperf/releases/...4.AppImage - if you would like to give it another try 16:16
16:18 MasterDuke joined
nine up and running :) 16:23
timotimo damn 16:24
that's pretty cool
i'm glad i spent the last three days agonizingly slowly inching towards this 16:25
now it has to be made plug-and-play for arbitrary raku modules / apps 16:26
and, well, rakudo itself without any special added stuff i guess :) 16:27
MasterDuke cool
timotimo github.com/timo/moarperf/releases/...4.AppImage - did you grab the link from backlog, MasterDuke?
MasterDuke i hadn't yet, thanks 16:28
nice, now i can run moarperf even when i've currently broken my rakudo (like right now, when my moarvm is on my MVMSpeshCandidate_as_REPR branch) 16:31
timotimo heck yeah
17:09 patrickb joined 18:41 patrickb left 19:00 AlexDaniel joined, AlexDaniel left, AlexDaniel joined 19:02 Altai-man joined 19:05 sena_kun left
Geth rakudo/rakuast: e0ced0d21b | (Elizabeth Mattijsen)++ | 3 files
Fixup Signature and Parameter deparsing

Plus some tweaks in other places
19:54
rakudo/rakuast: 0cd07cf00f | (Elizabeth Mattijsen)++ | t/12-rakuast/signature.t
Adapted tests for Signature/Parameter deparsing

Got stuck on handling slurpy, will revisit tomorrow
19:55 squashable6 left 19:56 squashable6 joined
lizmat afk& 19:58
19:59 squashable6 left 20:01 squashable6 joined 20:06 patrickb joined 20:16 patrickb left 21:29 Altai-man left