Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
AlexDaniel .seen usev6 04:45
yoleaux I haven't seen usev6 around.
AlexDaniel .seen bartolin 04:46
yoleaux I saw bartolin 9 Jul 2019 18:20Z in #perl6-dev: <bartolin> *RT tickets in the first sentence
AlexDaniel bartolin: duncand is wondering on the mailing list why not bump the minimum java version up to eleven 04:49
Geth rakudo: f1a7b9d68a | (Vadim Belman)++ | 3 files
Make CORE:: fully available at compile time.

Fix for rakudo/rakudo#3058		 05:35
rakudo: b455d2ec96 | (Vadim Belman)++ | 2 files
Re-consider what shall serve as CORE for 6.c

The outermost CORE seems to be doing the job well enough. Works with OO::Monitors.
rakudo/rakudo#3058
synopsebot RAKUDO#3058 [open]: github.com/rakudo/rakudo/issues/3058 [regression] PseudoStash rework breaks compile time aliasing
rakudo: c112ac12a6 | (Vadim Belman)++ (committed using GitHub Web editor) | 3 files
Merge pull request #3059 from vrurg/rakudo_3058

Make CORE:: fully available at compile time.
Fix for #3058
[Tux] Rakudo version 2019.07-61-gc112ac12a - MoarVM version 2019.07-9-ge07c0f252
csv-ip5xs0.672 - 0.689
csv-ip5xs-205.151 - 5.421
csv-parser23.467 - 23.757
csv-test-xs-200.432 - 0.440
test7.051 - 7.475
test-t1.921 - 1.956
test-t --race0.823 - 0.878
test-t-2030.058 - 30.193
test-t-20 --race9.114 - 9.980
 06:15
06:23 ufobat_ joined 06:26 ufobat__ left 07:04 patrickb joined 07:09 robertle joined
lizmat Files=1280, Tests=108463, 203 wallclock secs (26.93 usr 8.11 sys + 2813.71 cusr 286.42 csys = 3135.17 CPU) 07:32
yoleaux 06:11Z <jmerelo> lizmat: there's an issue for that. Not clear where, but I'll try to fix that ASAP
Geth rakudo: cygx++ created pull request #3061:
fix CLA URL		 07:54
rakudo: 9db4f847f8 | cygx++ | 2 files
fix CLA URL
 07:57
rakudo: f66c238ffa | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | 2 files
Merge pull request #3061 from cygx/fix-cla-url

fix CLA URL
patrickb .tell vrurg I have time this evening (/me in UTC+2). If you do to, we might finally manage to finish up #3031. 08:20
yoleaux patrickb: I'll pass your message to vrurg.
nine rakudo.org/downloads/nqp/ gives a 404. This breaks my packaging script 08:22
AlexDaniel: both, NQP's and Rakudo's Configure.pl now try to use git to update a submodule. This should not happen with a release .tar.gz. Otherwise compiling an older release will try to update the configure submodule with a high likelyhood of breaking. 08:56
Actually, lots of issues with builds and releases could be caught early by having a build on the Open Build Service become a part of the release process. 08:57
Aaand yet another issue: [ 134s] extracting debug info from /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/bin/perl6-debug-m 09:06
[ 134s] /usr/lib/rpm/find-debuginfo.sh: line 528: /dev/fd/62: No such file or directory
AlexDaniel :S 09:09
nine: can you file all that as tickets?
nine Where would I file an issue about rakudo.org? 09:12
Ah, found it. And noticed that there are actually already a couple of tickets and have been there untouched for 2 months 09:13
AlexDaniel yeah… 09:14
but personally I'm ok with github 09:15
even if we bring the downloads page back up, next time it breaks we'll have the same issues
nine I wonder what's so hard about adding a single redirect of that download page to github
AlexDaniel: with that line of reasoning, we shouldn't have a website at all, because it may break and cause issues 09:16
AlexDaniel nine: it's not going to unbreak your script though? :)
|Tux| Now that I am *not* working on that timing box, better timings:
Rakudo version 2019.07-63-gf66c238ff - MoarVM version 2019.07-9-ge07c0f252
csv-ip5xs0.727 - 0.747
csv-ip5xs-205.047 - 5.115
csv-parser21.822 - 22.433
csv-test-xs-200.422 - 0.431
test6.059 - 6.862
test-t1.621 - 1.685
test-t --race0.756 - 0.771
test-t-2028.017 - 28.142
test-t-20 --race8.781 - 9.020
AlexDaniel nine: well, to some extent, yeah. Most of the websites we had we shouldn't :) 09:17
nine Anyway, I wonder why appropriate tests are still not part of our release guidlines? 09:18
AlexDaniel there are a bunch of things tested and checked, just not the ones you're asking about right now :) 09:19
so you want no network access when building from a provided .tar.gz? 09:20
I mean, it's right, but what's the easiest way to test that? 09:21
nine Of course. What'd be the point of having tar balls otherwise?
Like I suggested earlier: use the Open Build Service
With a simple `osc build` I immediately got the errors: 09:22
[ 3s] Can't exec "git": No such file or directory at Configure.pl line 14.
[ 3s] Can't exec "git": No such file or directory at Configure.pl line 34.
[ 3s] Can't exec "git": No such file or directory at Configure.pl line 35.
lizmat is this something that can be fixed with 2019.08, or should we consider a 2019.07.1 ? 09:23
AlexDaniel I have no problem with 2019.07.1, though 2019.07 is not going to break if I understand things correctly 09:25
nine lizmat: the issue will hurt people who try to build older releases on machines with git installed. A point release wouldn't help there since the 2019.07 release will forever be affected.
AlexDaniel it just requires git and network access
nine AlexDaniel: no, the build actually runs without git (jay missing error checks!). I suspect it will break in the future with git installed since it will update the submodule to a newer and potentially backwards incompatible version 09:26
patrickb So there needs to be a way to block submodule updating in releases. 09:27
AlexDaniel wait, how come it's not pinned to a specific commit? 09:28
nine What does it use git for anyway? The calls are obviously not needed as everything proceeds just fine with them failing. 09:29
AlexDaniel yeah, it is pinned, of course 09:30
although… things are weird
so there's no git repo in the tar, so any action on the submodule is not going to work anyway, right? 09:32
nine I guess so 09:33
AlexDaniel and even if there is a git repo, it should still be pinned to 7411804dffa8d63c87956cc764727111537cd865 09:36
which also means that the last two commits in nqp-configure didn't make it into the release 09:37
is that right or am I missing something
so the only good reason for a point release is to get rid of annoying errors… 09:39
AlexDaniel afk for a bit 09:40
nine Oh, the strange find-debuginfo.sh error disappeared on a second try 09:44
But....there's new ones:
[ 120s] ERROR: RPATH "//usr/lib64" on /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/bin/perl6 is not allowed
[ 120s] ERROR: RPATH "//usr/lib64" on /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/bin/perl6-debug-m is not allowed
[ 120s] ERROR: RPATH "//usr/lib64" on /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/bin/perl6-debug is not allowed
[ 120s] ERROR: RPATH "//usr/lib64" on /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/bin/perl6-m is not allowed
[ 120s] ERROR: RPATH "//usr/lib64" on /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/share/perl6/runtime/dynext/libperl6_ops_moar.so is not allowed
patrickb: ^^^
patrickb nine: What does find-debuginfo.sh do? 09:58
nine #find-debuginfo.sh - automagically generate debug info and file list 09:59
#for inclusion in an rpm spec file.
patrickb nine: I'm in a rather bad position to debug this. find-debuginfo.sh seems to be a script part of the rpm build process, which I have never done before. I'd be grateful if you could do some digging yourself. I have time this evening and can help where possible. 10:12
Is looking for a .git folder a good way to check whether we are dealing with a release rakudo, i.e. no need for doing any submodule update stuff? 10:34
AlexDaniel patrickb: that's going to work, yes. It makes sense not to run submodule commands if there's no git folder 10:36
patrickb should be a two line change. I'll throw together a patch in the evening. 10:38
btw. I'll be at TPC as well. Looking forward to seeing some of you IRL. 10:39
patrickb is afk
AlexDaniel \o/
so if git is installed I get `fatal: not in a git directory` 10:41
but it doesn't affect the build at all, as far as I can see
lizmat: ok, I don't think there's any reason for a point release 10:42
but maybe we should be checking for stuff on stderr
nine patrickb: the find-debuginfo.sh thing seemed to be a one time only issue. The RPATH issue however persists 10:44
AlexDaniel well, unless there's something else :) 10:46
10:48 sivoais joined
Geth ¦ rakudo: AlexDaniel assigned to patzim Issue Configure.pl attempts to use git when it shouldn't github.com/rakudo/rakudo/issues/3062 10:51
nine Aaaaaand the next issue: [ 156s] error: File not found: /home/abuild/rpmbuild/BUILDROOT/rakudo-2019.07-2.1.x86_64/usr/share/nqp/lib/Perl6 11:17
AlexDaniel :S 11:19
nine No idea why that directory is gone
AlexDaniel “Rakudo NQP libs are now installed into /share/perl6/lib instead of /share/nqp/lib [34476c3a]” 11:22
maybe?
nine Yes, just found it in the ChangeLog
Finally.... 3 hours and 2 fixes later, what should be a 10 minute job is done :/ 11:36
Or not. WTF? [ 91s] Stage parse : New type Sub+{Callable[Nil]} for Sub is not a mixin type 11:37
build.opensuse.org/package/live_bu...5.0/x86_64
AlexDaniel [ 34s] + /usr/bin/gzip -dc /home/abuild/rpmbuild/SOURCES/rakudo-2019.03.1.tar.gz 11:40
why 2019.03.1 ?
so you're building rakudo 2019.03.1 on top of nqp 2019.07 11:41
nine Why indeed?
Aaaaah I was just looking too soon. 11:42
The rebuild had not been triggered for all targets after I pushed the update to the new version. Looks like the actual build is proceeding just fine: build.opensuse.org/package/live_bu...5.0/x86_64
AlexDaniel: thank you so much for pointing this out!
AlexDaniel nine: as for improving the release process, I'm all for it, but I'm a bit out of steam as a release manager. Please file some tickets pointing out what can be improved 11:44
nine: though note that the release guide does say that a produced tar must be tested, and the Sakefile follows that. I simply didn't see the error on stderr and honestly didn't really look because all steps passed fine 11:45
nine AlexDaniel: I wonder if we should actually take it a step further: automate packaging of every commit. That way the burden does not fall on the release manager but on the developer pushing
AlexDaniel nine: so I don't know if osc is going to help with that particular issue, but if it does help with anything else I'm all for it 11:46
nine: I'd also love spectests for every commit…
because this time 6.d-errata spectest didn't even work, and nobody noticed 11:48
well, nobody except the release manager who of course had to run it (and that is automated too, so can't make a release without doing it, but still)
it's like 20 spectests per day max, or so, not that much cpu time really 11:49
well, ok, not max but average :)
nine Even if....we actually have the computational ressources
lizmat github.com/rakudo/rakudo/issues/3063 :-( 11:50
AlexDaniel c: HEAD use NativeCall; 11:51
committable6 AlexDaniel, ¦HEAD(f66c238): «»
AlexDaniel lizmat: it's fine, if it's our fault I'll make a point release, no problem 11:54
then we can also fix some additional things :) 11:56
nine: fwiw github.com/nxadm/rakudo-pkg/pull/48 11:59
patrickb nine: I need more pointers wrt the RPATH thing. I was under the impression the lines you pasted were the output of `find-debuginfo.sh` as well. 12:00
12:00 lucasb joined
AlexDaniel 6c: (^∞).hyper.grep(*.is-prime)[1000]; now - ENTER now 12:00
committable6 AlexDaniel, gist.github.com/1dcccaf0be97fcb476...5e7f4eab1d 12:04
AlexDaniel 6c: (^∞).hyper.grep(*.is-prime)[1000]; say now - ENTER now
lizmat m: (^∞).hyper.grep(*.is-prime)[1000]; say now - ENTER now 12:05
camelia 0.51609888
committable6 AlexDaniel, gist.github.com/24b9c2bbeb48c61cac...4b0c0332ea 12:08
AlexDaniel c: 2019.03.1,2019.07,HEAD (^∞).grep(*.is-prime)[1000]; say now - ENTER now 12:10
committable6 AlexDaniel, ¦2019.03.1: «1.2757648␤» ¦2019.07: «1.244431␤» ¦HEAD(f66c238): «1.1550919␤»
AlexDaniel lizmat: this test is right on the edge github.com/rakudo/rakudo/blob/f66c....t#L33-L38 12:12
lizmat: so it's failing in travis sometimes
“This completes in .6s on 2018.03 but hangs on HEAD” 12:13
so it shouldn't be testing for 2x difference, 20x maybe?
if we want to test the hang, that is
and if the bug returns, it'll hang the whole test suite, right? :) 12:14
lizmat well, we haz the tools to a. make the time given different on version / release, and b. have it time out if it hangs 12:16
nine patrickb: MoarVM's Configure.pl doesn't set rpath in ldflags if $prefix eq '/usr' and rakudo must do so as well 12:24
patrickb nine: Got it. Thanks for the pointer! I'll see what I can do. 12:57
Geth roast: vrurg++ created pull request #559:
Add tests for CORE:: symbols availability		 13:02
roast: cba2d3b3b3 | (Vadim Belman)++ | 4 files
Add tests for CORE:: symbols availability

Tested scenarios are: compile time with combinations of EVAL and BEGIN.
rakudo/rakudo#3058		 13:03
synopsebot RAKUDO#3058 [open]: github.com/rakudo/rakudo/issues/3058 [regression] PseudoStash rework breaks compile time aliasing
Geth roast: 4581e617ea | (Vadim Belman)++ (committed using GitHub Web editor) | 4 files
Merge pull request #559 from vrurg/rakudo_3058

Add tests for CORE:: symbols availability
Tested scenarios are: compile time with combinations of EVAL and BEGIN.
rakudo/rakudo#3058
vrurg patrickb: As long as I'm currently done with CORE it should be possible today. 13:05
yoleaux 08:20Z <patrickb> vrurg: I have time this evening (/me in UTC+2). If you do to, we might finally manage to finish up #3031.
patrickb vrurg: Great! I'll be home in about three hours. 13:06
vrurg patrickb: with regard to R#3062 – I guess all is needed is to see if .git is there, right? 13:09
synopsebot R#3062 [open]: github.com/rakudo/rakudo/issues/3062 [BLOCKER] Configure.pl attempts to use git when it shouldn't 13:10
patrickb vrurg: Yes, my plan is to put `if ( -d '.git' ) {` around the code in `BEGIN{` at the top of Configure.pl. 13:11
vrurg Ok, thanks!
lizmat AlexDaniel: you're planning on a 2019.07.1 ? 13:14
re: github.com/MoarVM/MoarVM/commit/cfb5b3ae0f
fixing some issues with Cro in production :-) 13:15
vrurg lizmat: did you have a chance to test my fix? 13:22
lizmat vrurg: the one you merged ? yes 13:26
vrurg As I see no complains, the ticked could be closed? Right? ;)
lizmat ah, yes... :-) 13:27
vrurg Will do. Thank you!
lizmat beat you to it :-) 13:28
vrurg Arrgh! 13:30
14:18 lizmat left
Geth rakudo: vrurg++ created pull request #3064:
Remove release and codename properties from Compiler		 14:41
roast: vrurg++ created pull request #560:
Remove tests for release and codename on Compiler		 14:42
vrurg greppable6: codename 14:51
greppable6 vrurg, 3 lines, 3 modules: gist.github.com/9fa4650fecbba924da...fa2580140d
15:17 llfourn left 15:18 llfourn joined
jdv79 vrurg: noticed a comment you made about perf a couple months ago. its very much my #1 issue. i have a p5/p6 canary i use and its still almost 2 orders of mag diff (p6:p5:4900ms:60ms) 15:21
i have hit random bugs but except the async ones they were fairly minor 15:22
15:26 pamplemousse joined 15:41 robertle left 15:52 patrickb left 16:23 Kaiepi joined, lizmat joined 16:45 BeastieBot left, BeastieBot joined
AlexDaniel ok who has access to fix rakudo.org/files/rakudo ? 16:46
rba: ↑ ?
timotimo: ?
moritz: ?
16:49 lizmat left
AlexDaniel just add a redirect from rakudo.org/files/rakudo to github.com/rakudo/rakudo/releases 16:50
and then, what was the url for downloads
rakudo.org/dl/rakudo/rakudo-2019.07.tar.gz to github.com/rakudo/rakudo/releases/....07.tar.gz
and
timotimo are the release downloads on github not broken? 16:51
AlexDaniel rakudo.org/dl/rakudo/rakudo-2019.07.tar.gz.asc to github.com/rakudo/rakudo/releases/...tar.gz.asc
timotimo: I uploaded the same tar as I would have previously, so no, it is fine 16:52
do it as a pattern too, please. It wont work for older releases yet but I'll fix that shortly
rba: also, where's the list of all infrastructure things we have + notes on who has access? 16:53
also that dl/ stuff was probably just a redirect for 16:54
rakudo.org/downloads/rakudo/rakudo....07.tar.gz 16:55
16:57 patrickb joined 16:58 Kaiepi left
timotimo the first redirect stands 16:58
but i've gotta go afk now, maybe you can write the pattern for me in the mean time
gist.github.com/timo/231d72dfd701a...d27d5ad692 - the current config
16:59 robertle joined
AlexDaniel location ~ /files/nqp$ { return 302 github.com/perl6/nqp/releases; } 17:01
vrurg Who knows how to reset ChangeLog Draft to the initial state? 17:15
On the wiki, I mean.
Geth nqp: patzim++ created pull request #565:
Don't check for submodules in a release
rakudo: patzim++ created pull request #3066:
Don't check for submodules in a release		 17:16
AlexDaniel vrurg: done
vrurg AlexDaniel: thanks!
AlexDaniel timotimo: something like 17:20
location ~ /(dl|downloads)/rakudo/(rakudo-(\d\d\d\d\.\d\d(\.\d)?)\.tar\.gz(\.asc)?)$ { return 302 github.com/rakudo/rakudo/releases/...load/$3/$2 } 17:21
haven't really tested it, but the regex should match more or less correctly
17:37 lizmat joined
Geth rakudo: 78c05c1225 | (Patrick Böker)++ | Configure.pl
Don't check for submodules in a release

Releases already have all the necessary submodules included. Also we don't require git to be installed to build releases. Thus don't do any submodule checks.		 17:42
rakudo: 99e60a22c6 | (Patrick Böker)++ (committed using GitHub Web editor) | Configure.pl
Merge pull request #3066 from patzim/no-release-git

Don't check for submodules in a release
nqp: 2f1409adb8 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/templates/MOAR_REVISION
[MoarVM Bump] Brings 3 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...gab7bac5ad ab7bac5ad Avoid various races in profile/coverage//debug 2d54fa36b Fix dealloc profile logging cfb5b3ae0 Put limits on local and inline counts for inlining		 17:43
rakudo: c581be31d9 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/templates/NQP_REVISION
[NQP Bump] 2f1409adb [MoarVM Bump] Brings 3 co […]

NQP bump brought: github.com/perl6/nqp/compare/2019....g2f1409adb
roast: vrurg++ created pull request #561:
Document 6.e-related changes		 17:45
17:46 a3f left
Geth roast: a48d0c9581 | (Vadim Belman)++ | docs/announce/6.e.md
Document 6.e-related changes

Taken from rakudo/rakdup#3040		 17:46
roast: 23f73874bc | (Vadim Belman)++ (committed using GitHub Web editor) | docs/announce/6.e.md
Merge pull request #561 from vrurg/master

Document 6.e-related changes
17:46 lizmat left 17:50 a3f joined, a3f left, a3f joined
timotimo AlexDaniel: the pattern is in and the server restarted, please test it out 17:57
AlexDaniel timotimo: it does work. Same for nqp, please? 18:01
location ~ /(dl|downloads)/nqp/(nqp-(\d\d\d\d\.\d\d(\.\d)?)\.tar\.gz(\.asc)?)$ { return 302 github.com/perl6/nqp/releases/download/$3/$2 }
timotimo did only the "rakudo"s change to "nqp" and the "rakudo/rakudo" to "perl6/nqp"? 18:03
AlexDaniel yeah 18:04
timotimo should be active now
AlexDaniel nine: ↑ :) 18:08
yeah, it does seem to work
for older releases it currently 404-s but I'll try to fix that now… 18:09
vrurg Quick clarification from Moar experts: am I right that Decoder is about serialization/deseralization? 18:27
Geth nqp: 2c61a05a34 | (Patrick Böker)++ | Configure.pl
Don't check for submodules in a release

Releases already have all the necessary submodules included. Also we don't require git to be installed to build releases. Thus don't do any submodule checks.		 18:46
nqp: cf98e04cfc | (Patrick Böker)++ (committed using GitHub Web editor) | Configure.pl
Merge pull request #565 from patzim/no-release-git

Don't check for submodules in a release
AlexDaniel please add your commits to github.com/rakudo/rakudo/issues/3067 if you want them to be in the point release 18:49
jnthn: ↑ I assume some moar changes?
18:54 patrickz joined 18:55 Kaiepi joined 18:57 patrickb left
Geth rakudo: patzim++ created pull request #3068:
Don't use add any RPATH compile flags when installing to /usr		 18:58
rakudo: 1e0220ff59 | (Patrick Böker)++ | tools/lib/NQP/Config/Rakudo.pm
Don't use add any RPATH compile flags when installing to /usr

Do it just the way MoarVM also does it. The RPATH is not necessary in that case, as the libs will be in the library search anyways and having a
  rogue RPATH is bad style and trips up some packaging tools.
rakudo: 085ed2f572 | (Patrick Böker)++ (committed using GitHub Web editor) | tools/lib/NQP/Config/Rakudo.pm
Merge pull request #3068 from patzim/no-usr-rpath

Don't use add any RPATH compile flags when installing to /usr
AlexDaniel patrickz: why is /usr special? Are there any other paths where the same treatment should be used? 19:00
patrickz AlexDaniel: I think the idea is, that when installing to /usr it's a system wide installation. Then the lib would be located in /usr/lib and be in the library search path already. 19:01
AlexDaniel :S
patrickz The fix is pretty much copied directly from moar
AlexDaniel that sounds kinda fragile
patrickz Does!
19:02 MasterDuke joined
patrickz But fixing this better means turning changing the way moar has always done it. So for now I'd say we make sure it works as it worked in the past. 19:02
AlexDaniel okay 19:04
patrickz: so github.com/MoarVM/MoarVM/commit/f1...74b820eb39 is related, right?
patrickz Sort of related. But that change is irrelevant for rakudo as the rakudobuild does not create any libraries 19:06
AlexDaniel c: Bicycle say 42
committable6 AlexDaniel, ¦Bicycle: «No build for this commit»
19:14 lizmat joined
patrickz vrurg: Ping me, when you're available. 19:17
timotimo do we have some kind of guide to make a module that's based on a C library that i want to ship the source code with?
greppable6: MakefileFrom
greppable6 timotimo, Found nothing!
vrurg patrickz: ping, ping, ping? Is it ok? ;) 19:19
patrickz bonk bonk bonk 19:20
You could have aimed for anything but the head!
Geth ¦ problem-solving: rba self-assigned [WIP] perl6-infra overview github.com/perl6/problem-solving/issues/68
patrickz Time for another spaces-in-path try? 19:21
vrurg Are we about nqp today? 19:24
patrickz the spaces-in-path thing has changes in rakudo, nqp, nqp-configure and moarvm... 19:26
Or maybe I didn't get the question. 19:27
vrurg No, you got it right. :) 19:29
19:29 ufobat_ left
vrurg Sorry, just trying to multi-task. Might be asking wrong questions. 19:29
patrickz: did you branch your MoarVM? 19:32
patrickz what do you mean? github.com/patzim/MoarVM/tree/spaces-in-path <- that's the branch... 19:33
19:33 MasterDuke left
vrurg Ok, that exactly what I needed. 19:36
Another thing: it forcingly updates nqp-configure for me -> results in no read_config_from_command found. 19:37
It's when I try Configure.pl from rakudo and it goes into nqp. This is weird because you wouldn't be able to build it on your side. 19:38
timotimo greppable6: FromJSON 19:40
greppable6 timotimo, 32 lines, 5 modules: gist.github.com/e0c57c2199994626bd...7ef861bddb
patrickz Hm. I never tried building using --gen-nqp or so. I always built rakudo, nqp and moar separately and before calling ./Configure in nqp or rakudo I checked out the spaces-in-nqp branch in 3rd-party/nqp-configure
vrurg patrickz: I think you would need to pin the head of nqp-configure in your repo to nqp and rakudo repos for testing. Then reset it to perl6/nqp-configure after merge
patrickz: Then you miss the most commonly used scenario and this may result in bugs. 19:41
patrickz good point!
vrurg Tried to configure nqp separately. Somethow it gives me: You asked me to build MoarVM, but 2018.12-332-g5fa053ce9 is not new enough to satisfy version 2019.05-94-g459e68675 19:42
But MoarVM is checked out on space-in-path branch
patrickz does your moar repo have the original moar remote and all of it's branches and tags fetched? 19:45
There is some logic in there that retrieves that latest tag, which are only in the original rakudo repo...
Geth ¦ problem-solving: rba assigned to maettu Issue [WIP] perl6-infra overview github.com/perl6/problem-solving/issues/68
19:45 MasterDuke joined
vrurg patrickz: I build everything from your repos. MoarVM comes from there. build/mk-moar-pc.pl contains the 2018 version. 19:46
And yes, tags are outdated in your repo. 19:47
patrickz: I need to go now, will be back in ~2hrs. Mail me, please. 19:48
patrickz can you do `git remote add origin github.com/MoarVM/MoarVM.git && git fetch origin` in the moar checkout?
vrurg: OK
19:48 lizmat left
vrurg patrickz: I thought you were doing this to MoarVM too. Later. Thanks! 19:49
timotimo i would like to have method write-uint64(Buf:U: $value) and friends, please 19:55
also, default value for .read-uint64($offset = 0) and friends 19:56
is it better to have an rfc ticket or a pull request with an implementation? 19:58
[Coke] timotimo: is it controversial? 20:24
timotimo dunno
[Coke] if it's not hard to do, PRs make it easier to say yes, anyway 20:25
patrickz Where are the submodule versions set? 20:33
timotimo they are set by cd-ing into them, checking out a specific commit, then cd-ing back out and "git add"ing the submodule
i.e. they behave a little bit like a file, but you can cd into them 20:34
patrickz ugh
ok
vrurg patrickz: what do you do to build dyncall moar submodule 20:57
?
patrickz Hey 21:00
Did you try with a space in path? 21:01
dyncall is currently not able to do that on *nixes. I notified the devs. 21:02
So only without spaces on linux and mac
for now
vrurg I'm building in "~/tmp/rakudo test" and install into install/ under the path. 21:03
patrickz As I said, no spaces on mac and linux for now :-( 21:04
vrurg All of it doesn't make sense until we get it fixed. :( 21:05
patrickz It works with spaces on Windows. That's the platform that most dreadfully needs it.
I guess that's also the reason why the dyncall devs made it work on windows but not anywhere else. 21:06
vrurg Ok, built relocatable moar in a path without spaces, then renamed directory. That fooled it. 21:11
patrickz Not sure that means it worked or that it didn't... :-P 21:12
vrurg Half-by-half. Broke on including uv.h – because moar config reports the no-spaces path. 21:13
Fixed with a symlink.
Use of uninitialized value in subroutine entry at /Users/vrurg/tmp/rakudo test/tools/lib/NQP/Config/Rakudo.pm line 347. 21:14
patrickz that'd be nqp::prefix 21:15
vrurg And then lines 362,363 too. moar::ldrpath moar::ldrpath_relocatable 21:17
patrickz did it completely fail to retrieve the nqp config?
vrurg patrickz: Not sure, but it is likely so. ./install/bin/nqp-m --show-config gives all variables correctly. 21:19
21:22 lizmat joined
vrurg patrickz: I think I found the problem. It somehow reset nqp-configure to the default commit instead of space-in-path branch. 21:24
patrickz vrurg: I added a commit to rakudo and nqp that changes the nqp-configure submodule to point to my clone + latest commit. Did you see that? 21:25
vrurg patrickz: Yes, it did work for nqp but somehow rakudo did not update. Now everything built and installed (a little bit of dancing with tambourine can be ignored for now ;) ) 21:26
patrickz whoot! 21:27
So it's a thumbs up from the mac side?
vrurg So, just summarizing: works but needs cooperation from dyncall devs.
patrickz I notify them. They weren't aware of the problem but it doesn't have the highest priority at the moment. 21:28
vrurg patrickz: Confirm. If there any rough edges – we'll polish them later.
patrickz: Thanks! 21:29
patrickz OK. Then I'll revert the submodule test commit, rebase on master and merge. Thanks for your help!
vrurg is back to mangling with EVAL...
patrickz vrurg++
AlexDaniel patrickz: I don't like this: github.com/rakudo/rakudo/pull/3031...f0252622L3 21:41
patrickz: so I assume you changed it to do some debugging? 21:42
patrickz: but once you merge it to master, these revisions will be pointing to your repo forever
patrickz AlexDaniel: True. I just learned that's a good idea to ease testing (especially when asking others to test!). I'll remove the commit before merging. 21:43
AlexDaniel оч
ok
vrurg AlexDaniel: оч ok, aha... ;) 21:54
AlexDaniel vrurg: something got stuck in my mouth :) 21:57
22:10 Kaiepi left
MasterDuke .ask nine what is github.com/perl6/nqp/blob/master/mbc.nqp used for? 22:26
yoleaux MasterDuke: I'll pass your message to nine.
MasterDuke how does github.com/perl6/nqp/blob/master/s...erMAST.nqp know about $MVM_reg_num32 ? there's no `my $MVM_reg_num32` anywhere 22:32
timotimo lib/MAST/Nodes.nqp in moarvm has it 22:33
22:34 BeastieBot left
MasterDuke huh. any reason src/vm/moar/QAST/QASTRegexCompilerMAST.nqp doesn't use/include/whatever lib/MAST/Nodes.nqp ? 22:36
timotimo it doesn't have to 22:37
vrurg MasterDuke: quick guess: it's part of NQPHLL.setting?
timotimo it's probably gen-cat'ed in there 22:38
vrurg NQPCORE.setting, though
22:39 BeastieBot joined 22:54 MasterDuke left
AlexDaniel TreyHarris: found it! 23:40
23:40 patrickb joined
AlexDaniel my $resp = await Cro::HTTP::Client.patch: $url, :$headers, 23:40
:%body, content-type => ‘application/json’;
TreyHarris: perl6-mode doesn't get it and misindents the second 23:41
TreyHarris: github.com/perl6/perl6-mode/issues/21 23:43
23:44 patrickz left

Powered by Cro and the Raku® Programming Language.

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

Thank you!