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.
ugexe vrurg: i'm rerunning it for the 5th time now. its failed those same flaky tests every time. 00:20
Geth nqp: Altai-man++ created pull request #540:
Parse |, ||, &, && prefixes not in nibble		 00:39
rakudo: Altai-man++ created pull request #2907:
Parse `a & a` as a conjunction, not concat
roast: Altai-man++ created pull request #539:
Add test for GH #2901		 00:41
vrurg ugexe: I was away for all this time. Have like 2hrs to find out what's going on. 01:00
ugexe the tests that fail do things like "if it passes less than 16 tests then TODO the whole thing" 01:03
so its possible changes somewhere made it somehow pass enough tests to start failing 01:04
it would have to be in moar master 01:05
it could also just be incredibly flaky today
vrurg I need to compare what fails on appveyor to what fails on my win7 VM. Anyway, I did no changes that would affect tests. And yet... 01:08
ugexe the one that fails uses moar HEAD 01:14
not whatever moar-rev rakudo is currently pinned to 01:15
vrurg And I see that other PRs are failing same tests. Ok, at least I may no waste time on this and see what's wrong with OpenBSD build then... 01:16
ugexe i think its the performance test that is failing though
not those two TODOd tests
either way it wouldnt be from your commits 01:17
vrurg Though it puts me in doubts of wether to merge the PR now. Hope, Kaiepi would execuse me for that. :) 01:21
ugexe it looks related to some changes made in moarvm today actually 01:22
ci.appveyor.com/project/rakudo/rak...ta9j#L4681
thats the failure
vrurg BTW, I found the restart button – in a project I created for myself. Seemingly, appveyor doesn't think I have the right to restart a job.
Kaiepi there's a problem with the feed parallelization pullreq 01:25
after more benching it turns out to be the same speed or slower than the current implementation
ugexe do a bunch of blocking IO 01:26
i guess that isn't a good benchmark either 01:33
vrurg Not every task can be succesfully paralellized. The cost of inter-thread communication might be too high sometimes. 01:38
Kaiepi it's still slower even with blocking IO going on 01:45
MasterDuke why doesn't nqp have appveyor and circle-ci set up?
ugexe: does circle-ci offer BSDs? 01:46
vrurg Kaiepi: are you still available? 02:28
Kaiepi yeah 02:30
vrurg It's about your change to Process.pm6. You completely ignore nqp::execname() on OpenBSD. But it returns a valid value if moar started with --execname. 02:31
I'm gonna change it unless there is a reason not to do so. 02:32
Kaiepi that's alright 02:35
vrurg Kaiepi: Basically, it looked pretty reasonable before: nqp::execname() || $prefix ~ '/bin'. What was the point?
I checked and it returns an empty string without --execname. 02:36
Kaiepi it was returning random flags passed when i tested at one point 02:37
that may no longer be the case with the new moar runner
vrurg Ok, I'll give it a try. 02:38
MasterDuke vrurg: is there a reason rakudo's NQP_REVISION is in tools/templates/, but nqp's MOAR_REVISION is in tools/build/ ? 02:55
oh, did all your changes land already? or are some still waiting? 02:56
vrurg MasterDuke: I overlooked MOAR_REVISION. 02:57
MasterDuke ha, simple enough
vrurg Basically, eveything I needed I put in place. Just have no time for more changes. 02:58
So, fixing bugs is the priority for now.
MasterDuke sure, it's no big deal, just wondered if it was deliberate 02:59
vrurg The plan was to have build contain scripts and templates – data files of any kind. 03:00
Geth nqp: vrurg++ created pull request #541:
Move MOAR_REVISION into tools/templates folder where it belongs.		 03:28
nqp: b3c9cbe8fd | (Vadim Belman)++ | 5 files
Move MOAR_REVISION into tools/templates folder where it belongs.
 03:41
nqp: 5f5736c578 | (Vadim Belman)++ (committed using GitHub Web editor) | 5 files
Merge pull request #541 from vrurg/build-revamp

Move MOAR_REVISION into tools/templates folder where it belongs.
07:42 brrt joined 07:44 vrurg left 08:35 brrt left 08:37 brrt joined 09:01 brrt left 10:31 lizmat joined
MasterDuke jnthn, timotimo: why does github.com/perl6/nqp/pull/532 require the rebootstrap? 10:55
[Tux] Rakudo version 2019.03.1-398-g6a60a99fc - MoarVM version 2019.05-11-g248e2980a
csv-ip5xs0.744 - 0.841
csv-ip5xs-205.963 - 6.125
csv-parser23.392 - 24.064
csv-test-xs-200.435 - 0.467
test7.284 - 7.880
test-t1.714 - 1.734
test-t --race0.804 - 0.846
test-t-2028.862 - 29.035
test-t-20 --race9.430 - 10.093
 11:00
Geth rakudo: f18af94139 | Altai-man++ | src/Perl6/Grammar.nqp
Parse `a & a` as a conjunction, not concat

Needed to fix github.com/rakudo/rakudo/issues/2901
  jnthn++ for the actual fix		 11:01
rakudo: 1f89b12ee8 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/Perl6/Grammar.nqp
Merge pull request #2907 from Altai-man/master

Parse `a & a` as a conjunction, not concat
roast: 954d32b027 | Altai-man++ | S05-metasyntax/regex.t
Add test for GH #2901

  github.com/rakudo/rakudo/issues/2901
roast: b65646c8fd | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | S05-metasyntax/regex.t
Merge pull request #539 from Altai-man/test-conjunction-bug

Add test for GH #2901
timotimo MasterDuke: huh. i'm not sure, actually 11:04
lizmat touristing some more& 11:05
11:05 lizmat left
MasterDuke lizmat: touristing where? 11:08
timotimo: any reason not to merge as is? 11:11
timotimo can we find out how to get around the rebootstrap? 11:12
MasterDuke i would like to, but not sure how 11:17
timotimo without the rebootstrap it'll just not give you the options with their new behaviour? 11:18
MasterDuke in the help output, yeah
timotimo oh, the help output?
MasterDuke yep
timotimo hm, how is that built
MasterDuke oh wait, i did a `get reset --hard HEAD~1` and now it does? 11:19
timotimo ah, wonderful
perhaps just have to "make clean" a few extra times
MasterDuke timotimo: can you try my branch (at the commit before the rebootstrap) to be sure? 11:20
timotimo only nqp changes, right?
MasterDuke yeah
timotimo ugh, i had some extra commits on my master branch 11:24
huh 11:27
was nqp --help always like this? just a list of options? with their blah=o? syntax?
MasterDuke yep
timotimo oh, OK
maybe that does want a little bit of text 11:28
profile=s?
profile-compile=s?
profile-filename=s 11:29
profile-kind=s
that's correct, right?
MasterDuke yes. huh, it definitely wasn't before. oh well, guess i must have done something wrong 11:30
timotimo it's easy to do stuff wrong with our current build system :S 11:31
but someone's working on it!
MasterDuke i'll remove that rebootstrap and force push
timotimo cool 11:32
MasterDuke gonna merge unless there objections? 11:33
timotimo i don't think there's a reason not to merge it :) 11:34
Geth nqp/master: 5 commits pushed by (Daniel Green)++, MasterDuke17++
nqp: e49a70eb7e | Altai-man++ | src/QRegex/P6Regex/Grammar.nqp
Parse |, ||, &, && prefixes not in nibble

But in an appropriate regexes. It is needed for precedence like
  github.com/perl6/roast/blob/master.../rx.t#L828 to work,
which worked before due to a mis-parse.
  jnthn++ for the actual fix.		 11:51
nqp: 5276e899ae | Altai-man++ | t/qregex/rx_metachars
Fix test that opposed spec
nqp: 25114a3f13 | MasterDuke17++ (committed using GitHub Web editor) | 2 files
Merge pull request #540 from Altai-man/master

Parse |, ||, &, && prefixes not in nibble
rakudo: f077f57c08 | (Daniel Green)++ | tools/templates/NQP_REVISION
Bump NQP for regex grammar fixes
 11:57
12:45 dogbert17 joined
Geth roast: Altai-man++ created pull request #540:
Add test for GH #2602		 13:42
roast: Altai-man++ created pull request #541:
Add test for GH # 2480		 14:04
14:16 ggoebel joined
ggoebel fyi: rakudo.org/how-to-get-rakudo/ gives a 404 14:16
14:17 pamplemousse joined
timotimo aw, snap 14:18
that's probably one of the "redirects" that rakudo has set up for old URLs
hey ggoebel you don't happen to know both apache2 and nginx configuration? :) 14:19
github.com/perl6/rakudo.org/blob/m...pache.conf - this would want to be rewritten to nginx config format 14:21
more precisely, here's the nginx config we have at the moment 14:22
gist.github.com/timo/e9d36813a388d...e58714cb69
ggoebel ...and perl6.org/downloads/ is telling folks to download rakudo-star-2019.03.tar.gz
timotimo that seems to work 14:23
ggoebel: thanks for the report, perl6.org/downloads now points at rakudo.org/files/star/source instead of how-to-get-rakudo 14:28
MasterDuke hmm, if you --profile-stage=foo and want to set the filename, what should you use? --profile=filename, --profile-compile=filename, --profile-filename=filename (though the plan was to deprecate this option), or something else entirely? 14:36
timotimo i'd assume --profile-stage=optimize to go with --profile-compile=blerp.sql, and if you have --profile-stage without --profile-compile (i.e. with --profile=bloop.sql) it should perhaps warn? 14:40
14:47 pamplemousse left
MasterDuke do both --profile-compile=compile.sql and --profile=runtime.sql at the same time make sense? 14:47
15:02 pamplemousse joined
timotimo hmm 15:03
i mean, we can make it work
15:08 vrurg joined
MasterDuke right now it says `Profiling is already started` and then `Writing profiler output to compile.sql` 15:27
timotimo it might need a re-structuring of the call graph or something, dunno 15:30
MasterDuke so we should warn now about the conflicting options, but it could be made to work later? 15:32
ggoebel timotimo: thanks 15:34
16:36 vrurg left 16:37 vrurg joined 17:13 pamplemousse left 17:28 ufobat joined 17:32 ggoebel left 19:12 ufobat left 19:13 ufobat joined 19:37 ufobat left 21:58 lizmat joined 22:55 lizmat left, lizmat joined 23:05 lizmat left
Geth nqp: 125be965e3 | (Daniel Green)++ | src/HLL/Compiler.nqp
Disallow profiling both compile and run time

I.e., using --profile and --profile-(compile|stage) at the same time. This might be possible in the future, isn't now.		 23:53

Powered by Cro and the Raku® Programming Language.

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

Thank you!