00:04 vrurg joined
Geth_ nqp/truffle: 1bba032683 | (Daniel Green)++ | 2 files
[truffle] Fixes to build nqp-truffle-impl.jar
00:06
MasterDuke pmurias: fyi, i was following the substrate instructions, but the `native-image --language:nqp org.perl6.nqp.truffle.svm.Truffle6` step gives `Error: language:nqp contains unknown name in option specification: tool:truffleAvailable tool options are: tool:chromeinspector tool:profiler`. going to bed, hopefully will try to debug tomorrow 00:09
00:10 lucasb left 00:15 redable joined
MasterDuke pmurias: commenting out the `Requires = tool:truffle` in <graalvm_root>/jre/languages/nqp/native-image.properties got me farther 00:16
pmurias: `Error: Classes that should be initialized at run time got initialized during image building: org.perl6.nqp.truffle.NQPLanguage was unintentionally initialized at build time. To see why org.perl6.nqp.truffle.NQPLanguage got initialized use -H:+TraceClassInitializationError: Use -H:+ReportExceptionStackTraces to print stacktrace of
underlying exceptionError: Image build request failed with exit status 1`
pmurias: and the stacktrace i get if i add that option doesn't seem relevant 00:19
00:34 vrurg left 00:36 tony-o left 00:41 redable left
pmurias MasterDuke: doesn't ring any bells to me :( 00:42
00:44 squashable6 left 00:45 redable joined
pmurias MasterDuke: last time I was setting up substratevm I asked about on the graalvm/truffle gitter (now they moved it to slack) 00:45
00:45 squashable6 joined 00:49 vrurg joined
AlexDaniel vrurg: I see a bunch of merges of master into your branches, can you use rebasing instead? 00:58
00:59 pmurias left
vrurg AlexDaniel: I do so already. But sometimes it doesn't work. 01:00
AlexDaniel vrurg: it's not extremely important but it pollutes the git history unnecessarily, especially at the amount of work that you do :)
vrurg AlexDaniel: I know. Don't like it too.
What I had to learn at some point is to stash changes before rebase. 01:01
01:05 redable left
AlexDaniel vrurg: also, I advice against commit titles like “Fix for R#3035” 01:09
instead just say what you fixed
that's not your commit though :D 01:10
lizmat: ↑ :)
thing is, R#3035 is not even clickable
so even if I look at it through the web interface I'd still struggle
01:11 redable joined
vrurg Sometimes it's hard to shorten a bug report into a oneliner. Thus, some PRs get titles of the kind. But I definitely don't use R#. :) 01:12
AlexDaniel sure, but I'll end up shortening it into a nice title anyway once I get to changelogging, so… 01:15
vrurg Anyway, noted. Will be avoiding those. 01:16
AlexDaniel thanks! 01:17
vrurg np! 01:27
01:36 Kaiepi left 01:39 Kaiepi joined 01:42 redable left 01:45 redable joined 01:59 lucs joined 02:08 redable left 02:20 tony-o joined
Geth_ roast: c92b752668 | (Vadim Belman)++ | 3 files
Test for precompiled closures

In support of MoarMV/MoarVM#1209
Related to rakudo/rakudo#2897 and a number of other tickets.
03:42
roast: 54c942b7b1 | (Vadim Belman)++ (committed using GitHub Web editor) | 3 files
Merge pull request #593 from perl6/rakudo_2897_fix

Test for precompiled closures
05:05 squashable6 left 05:07 squashable6 joined 05:28 squashable6 left 05:30 squashable6 joined
nine .tell patrickb yes (assuming you mean a CompUnit::Repository::Installation): perl6 -I'inst#/path/to/folder' 06:36
tellable6 nine, I'll pass your message to patrickb
nine I never write issue numbers in my commit titles. They don't mean anything to the reader anyway. I usually try to mention the number in the commit message body though 06:41
07:43 Voldenet left 07:54 sena_kun joined 08:03 bartolin joined 08:36 bartolin left 08:37 bartolin joined 09:30 bartolin left 09:43 bartolin joined 09:55 Altai-man_ joined 09:57 sena_kun left 10:32 pmurias joined 11:08 squashable6 left 11:11 squashable6 joined
Geth_ nqp/truffle: 8b0e270b9f | (Paweł Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/runtime/NQPArguments.java
[truffle] Add some TruffleBoundaries

Don't inline in the HashMap implementation
11:28
nqp/truffle: 14259453ca | (Paweł Murias)++ | 2 files
[truffle] Add slow path specialization for method calls
11:35 Voldenet joined, Voldenet left, Voldenet joined 11:38 pmurias left 11:45 donaldh joined 11:56 sena_kun joined 11:58 Altai-man_ left 11:59 ggoebel joined 12:03 squashable6 left 12:04 pmurias joined 12:05 squashable6 joined 12:19 lucasb joined 12:23 ExtraCrispy joined 12:39 donaldh left 12:52 MasterDuke left 13:34 donaldh joined 13:56 Altai-man_ joined 13:58 sena_kun left 14:46 pmurias left 14:48 pmurias joined
[Coke] when I run "make spectest", what instance of perl6 is it using? one from the build dir, or the one from 'make install' ? 15:16
AlexDaniel [Coke]: it has to be the one from the build dir, if it's using one from install path then it's wrong 15:42
[Coke] AlexDaniel: danke. 15:43
just trying to test Inline::Perl5
ok. installed zef into that build dir. Installed Inline::Perl5 - make spectest still complaining that it's not installed. 15:44
$ install/bin/perl6 install/share/perl6/site/bin/zef install Inline::Perl5
(after installing zef into that dir)
make I screwed up the zef install.
*maybe 15:45
AlexDaniel oh, Inline::Perl5 tests, hmm 15:46
[Coke] finds the release guide instructions that he may have written at one point and tries that specifically. 15:47
AlexDaniel but these are based on doing `make install` 15:49
which is going to work for sure
well, it should, at least, but it doesn't make it more right? :) 15:50
nine [Coke]: that's a bug and patrickb is working on it 15:52
It is trivial to work around by changing t/harness5:45 15:53
[Coke] nine: oh
[Coke] picked a bad time to randomly run a test. :)
15:56 sena_kun joined 15:58 Altai-man_ left 16:01 patrickb joined
patrickb I just saw the discussion and thought I'd drop by. 16:01
tellable6 2019-10-29T06:36:58Z #raku-dev <nine> patrickb yes (assuming you mean a CompUnit::Repository::Installation): perl6 -I'inst#/path/to/folder'
patrickb The problem with Inline::Perl5 and the build runner is difficult. 16:02
nine It used to be simple? 16:03
patrickb It boils down to the question of whether we want the build dir raku to access outside repositories. 16:04
It worked in the past, because before I started refactoring the build dir perl6 was configured to access the installation dir repos.
So it also had access to all installed modules. 16:05
nine Sounds like a good thing?
patrickb Side effect: It created parts the repository folder structure during build (before install).
The way it currently is, the build runner has it's home dir (and with that the repo paths) set to a dummy folder in gen/ just to make sure we don't spam anything outside of the build directory before `make install`. 16:07
So it's a tradeoff. Either we make it access stuff outside the build dir, with the potential of writing there and the outside stuff messing up the build, or we forbid such access and not get automatic access to the outside repos. 16:09
nine So Inline::Perl5 can be installed into that folder. The harness can print instructions to do that.
patrickb That would be a way out for this specific use case. 16:10
nine AFAIK it's the only use case that's broken currently ;) 16:11
patrickb github.com/rakudo/rakudo/commit/cb...f61a596966 <- That's the commit that moved the build runner home dir.
Given we might actually want to use the home folder in gen/ we should rename it. Currently it's called "dummy_p6_home". :-P 16:12
16:12 pmurias left
patrickb [Coke]: You could try to install Inline::Perl5 into `$build_dir/gen/dummy_p6_home` and see if that makes the tests work again. I think it will. 16:14
nine: Do you think a solution that merely adds instructions to harness / harness6 be good enough for the Inline::Perl5 testing problem? 16:15
nine It would definitely be an improvement and allow for people to run those tests again. 16:16
patrickb nine: I'll look into this. 16:23
need to leave
16:23 patrickb left
vrurg When I'm done with a few tasks, I'm gonna open a ticket about changing build directory structure. 16:48
We must move things away from the base dir into a subdir. Most likely into blib.
16:53 samcv_ is now known as samcv 16:55 donaldh left 16:59 Kaiepi left
|Tux| Rakudo version 2019.07.1-451-g9f433da0d - MoarVM version 2019.07.1-309-g48491b896
csv-ip5xs0.709 - 0.720
csv-ip5xs-206.569 - 6.728
csv-parser21.434 - 21.795
csv-test-xs-200.424 - 0.428
test6.967 - 6.980
test-t1.763 - 1.820
test-t --race0.793 - 0.812
test-t-2030.184 - 30.598
test-t-20 --race9.184 - 9.274
17:19
17:27 MasterDuke joined 17:40 Kaiepi joined 17:55 Altai-man_ joined 17:57 sena_kun left
nine vrurg: is it possible that your context fix also fixed this todo in t/spec/S10-packages/precompilation.rakudo.moar? ok 23 - precompile exported cached sub # TODO RT #122896 18:29
vrurg nine: I think so. I'm getting somewhat strange behavior when run this test separately, perhaps due to incorrect handling of jvm todos. So, postponed looking closer at it. 18:31
TODO passed: 23 – this is what confuses me for now. 18:32
nine: Yep, it's a bug in harness5. Removal of todo from 23 just makes the test pass. 18:34
nine vrurg: I'm not sure it's a bug. It makes it very obvious when a todo passes, so we can remove the marker. 18:36
vrurg nine: The bug is about reporting 23 where just 1 is expected. 18:37
nine It's not "TODOs passed" :) It lists the test numbers that unexpectedly pass 18:38
vrurg is doing (facepalm) 18:40
AlexDaniel ehh I see a stresstest failing, I think 19:01
t/spec/S17-procasync/kill.rakudo.moar 19:02
rba: speaking of infrastructure-related things, there are two things that are currently painful to me 19:04
rba: one is that we don't run Blin regularly, another is that we don't run stresstests for every commit 19:05
rba: let me know what kind of beefy resources we have that I can fully abuse
rba: for Blin, more cores is better 19:06
stresstests I think I can even run on whateverable's server, but I'm not sure how to make it work… do we need to integrate it with github somehow, or something? 19:07
El_Che, jmerelo: what about creating a docker image for Blin, or something like that? 19:11
with all dependencies installed
here's the list of debian packages: github.com/perl6/ecosystem-unbitrot/wiki
19:12 jmerelo joined
jmerelo AlexDaniel: I could try, but I'm really swamped now with work. I might be a bit better from November 15th on... 19:12
AlexDaniel jmerelo: I'm in a similar situation, which is why I'm desperately looking for help :) 19:13
jmerelo AlexDaniel: tyil had his own set of Docker containers, built nightly...
AlexDaniel jmerelo: I don't need the docker container to be updated regularly
jmerelo AlexDaniel: I know, my point is that he's got the expertise to give it a try if he's not too busy with something else. 19:14
AlexDaniel tyil: ↑ ? help?
hm, I think even kawaii had something prepared for Blin… 19:15
github.com/perl6/Blin/pull/10
so I guess it just needs to be finished? 19:16
rba jmerelo: Found a copy/past error in authorized_keys for docs-stage.raku.org. 19:18
tellable6 2019-10-29T18:17:56Z #raku <jmerelo> rba you've got email
tyil currently busy, I'll take a look at the backlog to see what's wrong later 19:19
jmerelo rba: my bad. 19:20
rba: did you fix it?
rba jmerelo: Yes. Not your mistake.
jmerelo: Does your key work now? 19:21
jmerelo rba: let me try
rba: please note I added a new key, the one I use in my desktop
rba: and no, it's not working, still requesting the password
rba jmerelo: Saw it. Guessed the key called "OpenShift-Key" was the one I added last time. This might be yours too 19:22
jmerelo rba: no, wait, I'm using the wrong user
rba: let me try again
rba: right! Going! 19:23
rba: and here we go! docs-stage.raku.org/
rba++
rba: right, that's the one in my laptop
rba AlexDaniel: Not sure yet if we have a chance to get it, yet asked with jmerelos help for an AWS EC2 instance with 8 cores/32 GB. Yet calculate for a max usage of 3 hours/day. 19:24
jmerelo++
AlexDaniel rba: that's not enough
rba AlexDaniel: huhh.
AlexDaniel you can have 3 hours/day if you bump that to 24 cores or so 19:25
otherwise I think it'll take more time
jmerelo AlexDaniel: you mean for Blin?
AlexDaniel yea
jmerelo AlexDaniel: I thought we were only going to do a few compiles...
AlexDaniel: if we are going to run Blin, even if it's a few days a month, I guess it'll be a bit more... /cc +rba 19:27
AlexDaniel I'd say the goal for Blin is to run it every day, and for stresstest we should run it for every commit 19:28
so stresstest is easy, it's just a few minutes on any machine really
jmerelo AlexDaniel: OK, let me check then.
AlexDaniel the issue is more about how we're going to set it up, like where it's going to push the reports
but for Blin we actually need something relatively beefy 19:29
jmerelo AlexDaniel: best way is obviously to set up some Azure client that will spin a new instance only when we need it. Also, if possible, to use spot instances.
AlexDaniel: but I think that if we're set to run Blin every day, and for a good part of the day, we should use reserved instances. 19:30
AlexDaniel why? spot instances should be ok
I think
depends on how much cheaper they are
jmerelo Spot instances can be 90% cheaper, reserved instances 75% cheaper 19:31
19:31 El_Che joined
El_Che lo 19:31
AlexDaniel o/
jmerelo Only, well, they are reserved.
El_Che I am looking at Blin, it looks easy to create an image if you need one
AlexDaniel El_Che: see github.com/perl6/Blin/pull/10 19:32
rba jmerelo: As you where able to deploy to docs-stage.raku.org, you like to deploy too to docs.raku.org? 19:33
AlexDaniel maybe it just needs some brush-up
19:33 Xliff joined
Xliff \o 19:33
tellable6 2019-10-25T13:47:17Z #raku-dev <AlexDaniel> Xliff: oook… can you bisect it?
2019-10-25T13:48:38Z #raku-dev <AlexDaniel> Xliff: so it's between 2019.07.1-443-gcfc0f46c0 2019.07.1-398-gf3dda96a8, right?
AlexDaniel Xliff: o/ 19:34
El_Che yeah, that could work, however personnaly I would start from rakudo on the distro or rakudo-pkg if something newer is needed
AlexDaniel Xliff: ↑ these messages are a bit outdated, I tested it and found no significant performance regression
jmerelo rba: yes, please. Can you copy the last line in authorized_keys there? Thanks.
Xliff AlexDaniel: Ah. Saw your emails. Thanks for looking into it.
rba jmerelo: give me a sec.
AlexDaniel El_Che: I mean… can you do it? :) Feel free to do whatever you want to make it work
Xliff AlexDaniel: I think the problem is exacerbated by number of cores. 19:35
AlexDaniel El_Che: the issue with this image, IIRC, is that it's a bit tricky to get the files generated by blin back out of the image
El_Che Sure, I'll have it a quick go now, see how far I get
jmerelo AlexDaniel: pretty much everything for Blin happens in memory, right?
Xliff I will have to bisect on my own, but I am still involved in other things. I will try to get to it by the weekend.
El_Che ok, that's just a volume thing, isn't?
AlexDaniel El_Che: I… I guess so? I don't speak docker… 19:36
El_Che " -v /tmp/blin-output:/opt/perl6/blin/output:rw" citing kawaai in the ticket
ok, I'll cook something quick as a poc and we'll take it from there
AlexDaniel Xliff: before bisecting I'd recommend to first do a few runs to make sure there's an actual regression
Xliff Yes. 19:37
AlexDaniel jmerelo: not really, it does write to the disk like all the installed modules and stuff
rba jmerelo: try to depoly
deploy
AlexDaniel jmerelo: if you need it to be in RAM then just make it a tmpfs or something
Xliff AlexDaniel: I have the following versions 2019.07.1-273-g91d53ae7f and 2019.07.1-298-g67b6acf71
jmerelo AlexDaniel: some instances in AWS are memory-intensive, other CPU-intensive, other storage-intensive. Prices really vary. I was trying to find out which one would be the best 19:38
AlexDaniel IIRC rakudo memory usage improved quite a bit and you no longer need crazy amounts of RAM
jmerelo rba: docs.raku.org/ is live 19:39
AlexDaniel so more CPU, just a bit of storage (if any at all, doesn't matter), and it'll be fine
Xliff AlexDaniel: So I will test those versions, instead.
jmerelo AlexDaniel: it's not so much about the size, but the speed
Anyway
We're talking a factor of two or something like that here.
Xliff So, who can explain %?RESOURCES to me? ;)
Pretty please?
rba jmerelo: So, what are the next steps need to switch to docs.raku.org? 19:40
jmerelo Xliff: it's actually now included in the docs: docs.perl6.org/language/modules 19:41
rba jmerelo: I have made 301 redirects (permanent) with perl6.org/* -> raku.org/*
jmerelo rba: not for the documentation, please
rba jmerelo: That's why I ask. What is the idea to proceed?
jmerelo rba: it's in this document github.com/perl6/doc/issues/3046
And I really need to to AFK now. WRT AWS, if we use reserved instances and memory-intensive computing nodes it will be more or less in the ballpark 19:42
rba jmerelo: thanks for the initial deployment. 19:43
jmerelo rba: But if you want to add something so that it gets up to 1K, you can request double the instances we have now, and we'll allocate resources as we see fit
rba: impossible without your help... :-)
jmerelo goes off to dinner
19:44 ExtraCrispy left 19:45 jmerelo left
rba jmerelo: Not sure if I can now ask for the double. 19:45
tellable6 rba, I'll pass your message to jmerelo
tony-o filling up the db from p6c for the api 19:47
oops, not meant for this channel
Xliff AlexDaniel: "Use of Nil in string context". 19:52
AlexDaniel e: my $proc = Proc::Async.new: $*EXECUTABLE, "-e", "sleep 1"; my $prom = $proc.start; $proc.kill; await $prom; say ‘finished’;
evalable6 (exit code 1) The spawned command '/tmp/whateverable/rakudo-moar/9f433da0d67cfff5e0b8eacaf5d4c0abbf7c6ef4/bin/perl6' exited unsuccessfully (exit code: 0, signal: 1)
in block <unit> at /tmp/BmM_irnuQ5 line 1
Xliff Please note, This module is not currently installed. I thought using %?RESOURCES would handle both the "installed" and "not-installed" situations.
AlexDaniel c: 2019.07.1 my $proc = Proc::Async.new: $*EXECUTABLE, "-e", "sleep 1"; my $prom = $proc.start; $proc.kill; await $prom; say ‘finished’; 19:53
committable6 AlexDaniel, ¦2019.07.1: «finished␤»
AlexDaniel c: HEAD,HEAD,HEAD,HEAD,HEAD,HEAD,HEAD,HEAD,HEAD,HEAD,HEAD,HEAD my $proc = Proc::Async.new: $*EXECUTABLE, "-e", "sleep 1"; my $prom = $proc.start; $proc.kill; await $prom; say ‘finished’;
committable6 AlexDaniel, gist.github.com/d8fcb3b7529e7db4de...75e742d916
AlexDaniel bisect: old=2019.07.1 my $proc = Proc::Async.new: $*EXECUTABLE, "-e", "sleep 1"; my $prom = $proc.start; $proc.kill; await $prom; say ‘finished’;
bisectable6 AlexDaniel, Bisecting by exit code (old=2019.07.1 new=9f433da). Old exit code: 0
AlexDaniel, bisect log: gist.github.com/cbf4c86dc7504001f4...f3892612d2
AlexDaniel, (2019-09-05) github.com/rakudo/rakudo/commit/22...eda445227e
AlexDaniel 6c: my $proc = Proc::Async.new: $*EXECUTABLE, "-e", "sleep 1"; my $prom = $proc.start; $proc.kill; await $prom; say ‘finished’; 19:54
committable6 AlexDaniel, gist.github.com/9f91b4f67275f05c02...1589bc868c 19:55
AlexDaniel jnthn: what do you think?
nine Xliff: it does 19:56
AlexDaniel the test looks a bit wrong to me, but can we actually change that?
tyil ok
AlexDaniel: am I still needed?
nine Xliff: if you ask your actual question I may be able to answer it 19:57
19:57 Altai-man_ left 19:59 pmurias joined
rba AlexDaniel: May you let me know min/max for Cores/Memory/Disk-Space and how much GB over network per day or month? 19:59
AlexDaniel: Have the feeling that if we like to have a permanent solution, the cheapest would still be to get a used physical server. 20:00
AlexDaniel: Eg. www.ricardo.ch/de/a/hp-dl360p-gen8...074308174/ (sorry in German) and then ask someone to put it in there datacenter. 20:01
AlexDaniel rba: for memory 16G should be enough, maybe we can get away with less but 16 should be alright. Disk space doesn't matter, and more cores is better (it'll just run faster)
Xliff nine: Can you use %?RESOURCE data when a module is in an uninstalled state? 20:02
20:03 Kaiepi left
Xliff Because I have a library I've compiled and %?RESOURCES can't seem to locate it. I always get Nil, no matter what key I use. 20:03
And %?RESOURCES always seems to be empty
tony-o do you have the resources in your META Xliff ? 20:05
Xliff Yes
20:05 Kaiepi joined
AlexDaniel rba: what about those cloud services that give that stuff for free? For example, in the past I use a gce instance that had 24 cores, and they give any person free credit for one year 20:05
rba: so that's how I had trouble-free blinning for a year 20:06
Xliff cd glot.io/snippets/fhd17panto
AlexDaniel rba: but I used to run it just a few times per month, not every day
tony-o if you're using aws then a reserved instance is p inexpensive
rba You mean something like this? cloud.google.com/free/docs/gcp-free-tier 20:07
Xliff tony-o: That depends on the instance size.
tony-o i'm using the ECR stuff to run zef ecosystem testing (mirroring github + allowing users to upload/download directly)
if it's serving static content and running scripts in the background a medium instance is probably plenty, no? 20:08
Xliff tony-o: Yep. Should be.
AlexDaniel rba: yes. Also note that running Blin is not entirely secure, because there are modules that do network access and stuff, and I'd much rather allow these tests to run
tony-o i'll admit, i don't know what they're running
AlexDaniel rba: and by not entirely I mean completely not secure :) 20:09
Xliff Had a mediam instance just reserved (not running) and it was running me $20/month.
rba :-)
Xliff META6.snippet and directory list here -- glot.io/snippets/fhd17panto 20:10
20:10 donaldh joined
nine Xliff: need more info. That a full META6.json, directory tree and your invocation of the failing program would help 20:12
Xliff nine: The FULL META6.json is f(*&ing huge. I don't know why you'd need all of it. 20:13
nine What....is in your META6.json to make it huge? 20:14
Xliff GtkPlus
tony-o would also help: `find resources -type f`
Xliff That was included in the snippet! 20:15
rba El_Che: Like to go ahead with builing docker images. What base do you recommend? AlpineLinux? Or which one? Need both Raku and Perl5.
Xliff I'm loading up a GCP instance. 20:18
20:30 donaldh left
El_Che older zef's are broken it seem 20:31
Xliff 82 seconds to compile CORE.c -- not impressive on the standard setup. 20:32
El_Che rba: will get back to you, I'll probably release new pkgs with a newer zef, but that won't fix zef in $HOME is the breakage is changes in the ecosystem
Xliff tony-o, nine: Updated snippet glot.io/snippets/fhd1xnu2lz 20:33
I had to remember my login so I can save. :P
nine Xliff: try perl6 -I. instead
Xliff: oh, %?RESOURCES is for modules
it won't work in a -e like that 20:34
Xliff So it has to be in a .pm6 or can it be a .pl6
nine It has to be in something that's loaded by a `use`, `require` or `need` statement 20:35
rba El_Che: What is broken in the older zef? I've troulble-shooted ecosystems-api.p6c.or already.
nine Well actually just `use` and `need` as %?RESOURCES is compile time
rba El_Che: Wanted to make sure nothing old is broken. There might be many installations around with old zef.
El_Che I had 0.7.8 and I had to nuke it and install 0.8.1
"Method 'content' must be implemented by Zef::Distribution because it is required by roles: Distribution."
it also happens on a fresh container so I don't think it's me 20:36
Xliff nine: Well something is happening. I got "Invalid JSON"
El_Che I'll bump my packages to 0.8.1
20:37 pmurias left
El_Che 22 distro/releases so it will take a while 20:37
sadly 2019.07 is a bitch on i386
9/10 test fail on travis, because the timing is to small. It should be fixed on the upcomng rakudo release
rba El_Che: Doesn't sound like what we had. Anyway was older, was 0.7.1 I guess... 20:38
20:38 pmurias joined
El_Che 0.7.4 20:39
github.com/nxadm/rakudo-pkg/blob/1...vis.yml#L4
bumped to 0.8.1 and creating a updated pkg
MasterDuke pmurias: built a native image of the point benchmark and can run it, but it's even slower 20:41
Xliff How can I convert "2019.07.1-273-g91d53ae7f" to a commit hash? 20:42
nine Xliff: take everything after the -g
Xliff Ah. 20:43
I knew that.
Xliff is not having a swell day./
El_Che rba: anyway, it's a good excuse to introduce the new fedora 31, Ubuntu 19.10 and CentOS 8 packages :)
rba El_Che++ 20:44
El_Che: So which base? Alpine?
El_Che AlexDaniel: a poc Blin img should be up soon once the rakudo-pkgs are built with the new zef
rba: for Blin, debian as documented
rba: it's a blin image, not a generic rakudon one 20:45
rba El_Che: and for the website raku.org?
El_Che rba: but creating a generic rakudo image is as simple and adding a FROM <some distro> and adding the rakudo-pkg repo
rba El_Che: blin image I will try to spin on a server maettu has around...
20:46 pmurias left
AlexDaniel El_Che: if you can also push the dockerfile or something to perl6/Blin, possibly with the docs on how to get it going, that'd be awesome 20:46
El_Che sure, let's see how to poc goes and then you're free to put it whatever you want (I'll transder the repo)
20:47 pmurias joined, redable joined
El_Che github.com/nxadm/raku-blin-docker 20:47
wating on new rakudo-pkg to be sure it work 20:48
but the idea is to just run : dblin $what_options_and_parameters_you_give_to_blin.p6
rba afk
El_Che output will be in /var/tmp/blin-output 20:50
20:58 redable left
El_Che should have a look at github action: travis hasn't been great the last months... 21:02
AlexDaniel El_Che: maybe do /var/tmp/blin:/Blin so that the whole directory is shared? 21:04
El_Che sure
AlexDaniel there's also a data folder and other stuff that could be useful
21:08 redable joined
MasterDuke pmurias: but it got a bunch faster and stopped spewing exceptions when i commented out this line github.com/perl6/nqp/blob/truffle/...e.java#L30 21:11
Geth_ nqp/truffle: 0b3a4cb89c | (Daniel Green)++ | 2 files
[truffle] Fixes for native image building
21:26
pmurias MasterDuke: native images are slower 21:33
MasterDuke: the ones built with the CE version instead of the EE ones have as far as I know (some) JITing disabled 21:34
MasterDuke ah
Xliff 91d53ae7f <- this is invalid, but I know that was the hash after perl6 2019.07.1-273
Trying to bisect something, now.
Should identify with a commit made on 20191015 21:35
Can someone help?
pmurias MasterDuke: the graalvm team doesn't seem to like talking how the EE version is different tho 21:43
MasterDuke heh, why am i not surprised
pmurias if this means they put more features in the open source version that way and market the EE version as being superior in an "unspecified way" I'm happy with the arrangement 21:46
MasterDuke sure, no skin off my back 21:48
Xliff: that's a moarvm commit 21:50
Xliff: github.com/MoarVM/MoarVM/commit/91d53ae7f
Xliff Ah. 21:51
MasterDuke pmurias: any suggestions for easy stuff i could do next? 21:55
pmurias t/nqp/019-file-ops.t contains a lot of stuff that seems fairly easy to do 21:58
OTOH not sure how fun of a task that would be 21:59
22:04 redable left
MasterDuke oh, i had been looking in there, forgot about that. i'll take a look (probably not this evening though, think i'm going to go play some disgaea 5 for a bit) 22:06
22:09 pmurias left 22:13 epony left 22:25 pmurias joined
pmurias MasterDuke: I'll try to think of something cooler to do 22:37
23:51 epony joined