AlexDaniel but then you probably want it to explode? 00:00
because most likely it didn't fully finish successfuly
jnthn Yes, but if it didn't, you could have accidentally relied on it not doing so...somehow.
But given this was only a problem for certain ways of it getting killed, and normal unsuccessful exits already exploded, I suspect it will not hurt so much. 00:01
AlexDaniel if this breaks my own code I'd be so pissed :D 00:02
pmurias jnthn: I get 3.5 for moar, 14s for truffle and 35s for the jvm 00:06
interesting why MasterDuke and mine jvm times are that different
MasterDuke pmurias: oh, i bet i'm still using the graal jdk/jre to run the old code 00:07
hm, nope, don't think so 00:08
Xliff Rig diffs. 00:10
pmurias MasterDuke: I'm using the graalvm-ce-19.2.1 for both
Xliff As in... what are your respective machine specs...
MasterDuke openjdk version "13.0.1" 2019-10-15 OpenJDK Runtime Environment (build 13.0.1+9) OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode)
for regular code
OpenJDK 64-Bit GraalVM CE 19.2.1 (build 25.232-b07-jvmci-19.2-b03, mixed mode) for truffle 00:11
and ryzen 3700x is the hardware 00:12
pmurias MasterDuke: have you compared running the old backend with the GraalVM CE (just using a regular ./nqp-j with the same java)
MasterDuke pmurias: i haven't. i'll do that tomorrow, but i'll have to rebuild it (i just tried without and got `Error: A JNI error has occurred, please check your installation and try againException in thread "main" java.lang.UnsupportedClassVersionError: org/perl6/nqp/runtime/CompilationUnit has been compiled by a more recent version of the Java Runtime 00:14
(class file version 53.0), this version of the Java Runtime only recognizes class file versions up to 52.0`
pmurias hmm, seem super weird 00:15
Xliff Would someone appreciate another benchmark? 00:16
I could run one, if someone told me how.
$ perl6 scripts/dependencies.pl6 -> 20
Oops 00:17
pmurias MasterDuke: you get that error for the nqp-j in a non-truffle checkout?
Xliff perl6 -e '$*KERNEL.cpu-cores.say' # 20
MasterDuke pmurias: yeah, i just put the graal directories at the beginning of my PATH (like they are when i run the truffle code) 00:18
00:18 tyil[m] left
pmurias Xliff: I have to hit the bed asap as I'm not thinking straigh, stayed up too late to see how the optimized WVal will work 00:19
00:19 tyil[m] joined
MasterDuke Xliff: nqp code is `class Point { has $!x; has $!y; method new(:$x, :$y) { my $obj := nqp::create(self); nqp::bindattr($obj, Point, q|$!x|, $x); nqp::bindattr($obj, Point, q|$!y|, $y); $obj; }; method x() { $!x }; method y() { $!y } }; my $total := 0; my int $i := -1; while ++$i < 50000000 { my $p := => 2, y => 3); $total := $total + 00:19
$p.x + $p.y; }; say($total)`
how long does that take with the jvm backend for you? 00:20
but i too need to get to bed
Xliff MasterDuke: I don't think I build the jvm backend. Let me get that running. I will .tell you and pmurias the results.
pmurias jnthn: the truffle backend still doesn't have method call optimized properly (with polymorphic inline caching) one I figure out how to do it in a way that's actually faster the benchmarks should be more representative 00:22
MasterDuke Xliff: this always bites me. if you get weird errors during build, make sure you're not parallelizing your make call (i.e., `make install`, not `make -j16 install`)
Xliff MasterDuke++ # Thanks on the heads up! 00:23
"javac: invalid flag: --release" ??
I'm using openjdk-8-jdk
MasterDuke -version
Xliff Feh. Now installing -14 00:24
00:24 tbrowder joined
Xliff Ah. There we go. 00:27
00:30 pmurias left
Xliff Exception in thread "Thread-4" org.perl6.nqp.runtime.UnwindException 00:36
'/home/home/cbwood/Projects/rakudobrew/versions/jvm-2019.07.1/perl6-j' '/home/home/cbwood/Projects/rakudobrew/versions/jvm-2019.07.1/tools/build/install-core-dist.p6' '/home/home/cbwood/Projects/rakudobrew/versions/jvm-2019.07.1/install/share/perl6'
I'm going to leave the build running, but the process seems hung at this point with it repeating that exception message. 00:42
MasterDuke, pmurias: Aborting jvm make... something isn't write either with my system or the build process. It is hanging at: '/home/home/cbwood/Projects/rakudobrew/versions/jvm-2019.07.1/perl6-j' '/home/home/cbwood/Projects/rakudobrew/versions/jvm-2019.07.1/tools/build/install-core-dist.p6' 01:02
01:09 ggoebel joined 01:14 Xliff left 01:37 ggoebel left
tyil I think I got it working 01:42
vrurg tyil++ 01:51
Geth_ roast/rakudo_3257: 931d803d7c | (Vadim Belman)++ | 2 files
Consider newly added X::SymbolNotDynamic exception

Make _S32/exceptions/mist.t_ use v6.c explicitly
roast/rakudo_3257: 5074c079e1 | (Vadim Belman)++ | S02-names/pseudo-6e.t
Additional tests for dynamic pseudo-packages

Support for rakudo/rakudo#3272
roast: vrurg++ created pull request #595:
Additional tests for dynamic pseudo-packages
04:49 ggoebel joined 05:03 b2gills left 05:05 b2gills joined 05:33 AlexDaniel left 06:38 jmerelo joined 07:31 ggoebel left 07:33 jmerelo left 08:08 sena_kun joined 09:15 ExtraCrispy joined
El_Che AlexDaniel`: I'll rebuild the image 09:19
09:44 Altai-man_ joined 09:46 sena_kun left
El_Che AlexDaniel`: the fix seem to work. Uploading the new image 10:00
AlexDaniel`: upload done. Test it, see if it's ok, then we'll see what to do with it 10:08
10:13 AlexDaniel joined, AlexDaniel left, AlexDaniel joined
AlexDaniel El_Che: I haven't slept in a lot of time, so going to bed now. Will test it once I wake up. Alternatively you can run the example you have in the readme 11:11
this one: ./dblin -old=2018.06 --new=2018.09 Foo::Regressed Foo::Regressed::Very Foo::Dependencies::B-on-A
if it bisects both regressed modules to v2018.08.48.g741ae6f4e then it totally works
El_Che Foo:: is a thing :) 11:12
AlexDaniel you'll see the result in the output folder
El_Che AlexDaniel: no rush, about it, just telling it worked for me
AlexDaniel okay, well
11:12 ExtraCrispy left
El_Che AlexDaniel: I run it for all inadvertingly 11:12
and wondered why my pc was so slow :) 11:13
AlexDaniel hehe
El_Che will check with the modules you gave me
AlexDaniel now we need to figure out where we can actually run it
because I still want to see the result before we ship this release
El_Che does it take than 50 minutes to run and it shows some output along the process? ==> travis 11:14
otherwise we could try to get some aws credits for foss?
AlexDaniel iirc it takes like two hours if you have 24 gce-like cores 11:15
El_Che so, travis is out
I have a account, let me check if the limit apply there
limit is there 2h 11:20
AlexDaniel: sleep tight, Blin is happely running here. It looks good 11:21
AlexDaniel niiice 11:23
El_Che: but it won't make it in 2 hours
El_Che will run it in a vm on a other machine, because I need this one to work :) 11:30
tyil I have some minor tweaking to do in order to feel confident about this rakudo-star 2019.07.1 build, but for future reference, how does one make a possible r* build "official"
(fwiw, latest build over yonder:
also, is there a place I can find general information about the release schedules of NQP, MoarVM and Rakudo? 11:34
or RSS feeds with their releases
AlexDaniel tyil: there's [email@hidden.address] where releases are announced 11:37
I mean, the mailing list
tyil: NQP and Rakudo are usually released at the same time and the announcement is for both 11:38
tyil: as for schedules, it's currently “when it's ready” until someone better than me takes over :) 11:39
tyil: as for making it official, last time we had this ticket: 11:40
so clarkema and hankache volunteered but I don't know the current situation and what they're up to 11:41
see this ticket also:
if you figured it all out, perhaps you should let others know
tyil: did I answer any of your questions? :) 11:42
I don't feel like I did
tyil: also, I've seen your question about nqp makefile, and I'm a bit confused
these docs describe all steps required to make a release, including when to run 11:43
El_Che I think tyil is talking about Rakudo Star and not Rakudo. They are not related.
AlexDaniel I know, but previous question was about nqp, I think 11:44
El_Che ok 11:45
AlexDaniel well, star has a file for that too
tyil AlexDaniel: about NQPs release guide, step 3 says it's VERY important to ensure everything is clean, and tells you to do a make realclean
AlexDaniel but I've never tried following the steps there
11:45 sena_kun joined
tyil but you can't do a make realclean there 11:45
because the makefile is generated in the next step
as for the ML, I'll subscribe there
and I'll also post on the github issue to inform people I do have a working build environment
AlexDaniel tyil: yeah, it assumes that you were already using the repo to do stuff, which is a horrible assumption really :)
tyil I can make a PR on the GitHub repo is needed for it 11:46
11:46 Altai-man_ left
AlexDaniel for rakudo and nqp releases I use this: 11:46
I really wanted to make the code there nice, but… I guess it works totally fine, so why touch it? 11:47
tyil I don't want to depend on Perl 6 to build/release these things, because it feels wrong to depend on the thing you're building :p
but that's a personal opinion, if that works for you it's all good ofc
AlexDaniel well, you can follow the steps manually 11:48
11:48 MasterDuke left
AlexDaniel which is even worse because many times it was proven that people do mistakes :) 11:49
tyil I try to stick to POSIX shell where possible, but R* was doing Makefile stuff already so I mostly just altered that to get things working again
AlexDaniel: is perl6-compiler a private list? I don't see it on 11:50
El_Che tyil: are you coordinater to the star release managers (in order to avoid doing dubble work)?
tyil El_Che: no, I was just bored at work and wanted a newer build
AlexDaniel tyil: see 11:51
tyil thanks!
El_Che tyil: ping them, maybe you fixed what was blocking them to do a new release
AlexDaniel: ⏳ 134 out of 1343 modules processed 11:52
this will take a while :)
AlexDaniel also, that sakefile is not for building, it's for releasing
so there's no issue depending on perl6 itself, as far as I can see
I mean, raku, right 11:53
El_Che AlexDaniel: the volume is there, the output file also, so I think the image does what it's supposed to do
AlexDaniel El_Che: yeah, we'll see later
El_Che AlexDaniel: now go to bed :)
AlexDaniel thanks :) 11:54
13:03 ggoebel joined 13:08 ZzZombo_ joined 13:10 ZzZombo left, ZzZombo_ is now known as ZzZombo 13:44 Altai-man_ joined 13:46 sena_kun left, Altai-man_ left 13:47 sena_kun joined 14:49 ExtraCrispy joined
Guest13443 releasable: next 14:54
releasable6 Guest13443, Next release will happen when it's ready. There are no known blockers. 248 out of 451 commits logged (⚠ 9 warnings)
Guest13443, Details:
Guest13443 releasable: prev
releasable6 Guest13443, I cannot recognize this command. See wiki for some examples:
15:12 patrickb joined
[Tux] Rakudo version 2019.07.1-451-g9f433da0d - MoarVM version 2019.07.1-311-g8caf9fd58
csv-ip5xs0.725 - 0.744
csv-ip5xs-206.677 - 7.121
csv-parser23.838 - 24.244
csv-test-xs-200.432 - 0.442
test7.378 - 7.638
test-t1.741 - 1.788
test-t --race0.854 - 0.868
test-t-2030.764 - 31.108
test-t-20 --race9.621 - 10.147
15:43 Altai-man_ joined 15:46 sena_kun left 15:57 japhb left 16:27 Guest13443 left
tbrowder hi, folks, i'm trying to find where nqp::hllize is defined and what it does. i can find hllize_nfa but not hllize. can anyone help? thnx 16:35
16:43 jmerelo joined
vrurg tbrowder: look in MoarVM src/core/interp.c 16:46
tbrowder vrug: thanks! 16:48
vrurg tbrowder: basically, it boxes a type into what current compiler language understands. I.e. native int -> Int, etc. But I don't know details. 16:50
16:51 patrickb left 17:16 japhb joined 17:45 sena_kun joined 17:46 Altai-man_ left 17:58 jmerelo left 18:20 lucasb joined
tyil oh, GitHub now also has a docker registry for repos 18:21
wonder if I can make my gitlab stuff for r* work there too
El_Che .tell jmerelo: the doc packages should be up (updated to 2019.07.1) 18:26
tellable6 El_Che, and I oop! Backtrace:
El_Che tell jmerelo: the doc packages should be up (updated to 2019.07.1)
.tell jmerelo the doc packages should be up (updated to 2019.07.1)
tellable6 El_Che, and I oop! Backtrace:
El_Che right, I broke tellable6 now
"Couldn't decode hexadecimal unicode escape \uD83D\uDE47‍♂️ at 606925" <-- WHAT? 18:27
tobs maybe it is related to the aptly named FootgunDB a few frames up? :D 18:28
El_Che tobs: missed it 18:30
AlexDaniel El_Che: that's a bug in JSON::Fast
it can't roundtrip some unicode 18:31
18:31 tellable6 left
El_Che I don't get the maile unicode thing 18:31
is not in my msg
AlexDaniel yep, its in its db
El_Che lol
AlexDaniel so somebody said something with unicode, it saved it, now it can't decode it back 18:32
it's a pretty serious issue in json::fast IMO
18:34 tellable6 joined
AlexDaniel El_Che: try again? 18:34
El_Che .tell jmerelo the doc packages should be up (updated to 2019.07.1)
tellable6 El_Che, I'll pass your message to jmerelo
El_Che \o/
AlexDaniel El_Che: how's blin? 18:36
El_Che it keep runnig fine for a few hours then I had to stop it because I ran of space 18:37
looks good to me
output is there, tests run fine, etc
18:38 pmurias joined
AlexDaniel El_Che: out of space?? How much space did it take? 18:42
El_Che it was a small vm
the img took quite a lot and then all the run time modules it was testing
around 4gb
19:30 MasterDuke joined 19:43 Altai-man_ joined 19:46 sena_kun left
MasterDuke pmurias: i can't build the regular jvm backend with graal because graal doesn't know java.lang.ProcessHandle. we require java 9, so we'll have to wait for graal to update 20:10
pmurias MasterDuke: when you build the truffle backend ./nqp-j (vs ./nqp-j nqp-truffle.nqp) gives you a nqp that's almost the same as the regular jvm backend from HEAD 20:43
MasterDuke right. then for 10000000 i get 3s with truffle and 6.4s with just nqp-j 20:45
pmurias MasterDuke: that's with counting in the startup? 20:53
MasterDuke yeah, just using `time`
20:53 ExtraCrispy left
MasterDuke pmurias: do you know how to use jvisualvm (which comes with graal) to profile nqp? i've gotten it running and attached to the nqp process, but i can't seem to get a profile information generated 20:55
20:58 pmurias left, pmurias joined 21:34 Altai-man_ left
MasterDuke the github description for rakudo says 'Raku on MoarVM and the JVM'. should we add JS to that? 21:35
AlexDaniel yes 21:38
like this?
MasterDuke sure. i prefer the oxford comma, but it's not technically required 21:42
21:50 lucasb left
AlexDaniel sure! 21:51
now people will complain about an unnecessary comma :) 21:52
El_Che People like me :) 22:08
It feels as redundant as ";" at the end of raku statements :) 22:09
lizmat } 22:11
[Coke] oxford++ 22:51
AlexDaniel jnthn: cro release? 22:59 23:00
23:00 pmurias left 23:02 pmurias joined
jnthn AlexDaniel: Probably tomorrow. Today was Comma release. :) 23:10