00:07 patrickz joined 00:11 patrickb left 00:42 sena_kun left 00:43 lucasb left 00:44 patrickz left
Geth rakudo: 9eef0e63af | (Rod Taylor)++ | src/core.c/DateTime.pm6
Allow DateTime to accept picoseconds

This was restricted via regex and can be enlarged with minimal change. Output remains at 6 digit precision, with the remainder rounded away by the default formatter. This additional information may be obtained with the second function or by providing a custom formatter.
my $dt = DateTime.new(:2010year, :10month, :10day, :5hour, :5minute, second => '59.987654321'); say $dt; # output 2010-10-10T05:05:59.987654Z say $dt.second; # output 59.987654321
rakudo: abe287cf8e | (Vadim Belman)++ (committed using GitHub Web editor) | src/core.c/DateTime.pm6
Merge pull request #3396 from rbt/master

Allow DateTime to nanoseconds.
vrurg notable6: blogs.perl.org/users/vadim_belman/2...essor.html 00:51
notable6 vrurg, I cannot recognize this command. See wiki for some examples: github.com/perl6/whateverable/wiki/Notable
vrurg notable6: weekly blogs.perl.org/users/vadim_belman/2...essor.html 00:52
notable6 vrurg, Noted! (weekly)
00:57 sena_kun joined 02:01 unicodable6 left, sourceable6 left, statisfiable6 left, reportable6 left, coverable6 left, notable6 left, quotable6 left, bloatable6 left, benchable6 left, releasable6 left, greppable6 left, squashable6 left, nativecallable6 left, bisectable6 left, shareable6 left, committable6 left 02:02 reportable6 joined, bisectable6 joined, sourceable6 joined, coverable6 joined, shareable6 joined, nativecallable6 joined 02:03 squashable6 joined, committable6 joined, notable6 joined 02:04 benchable6 joined, greppable6 joined, statisfiable6 joined, bloatable6 joined, unicodable6 joined, quotable6 joined, releasable6 joined
Geth roast: 5b211db7be | (Rod Taylor)++ | S32-temporal/DateTime.t
Tests for second truncation to milliseconds

This was already being done by the formatter. Add tests to ensure the behaviour remains consistent.
roast: 17407a3f47 | (Vadim Belman)++ (committed using GitHub Web editor) | S32-temporal/DateTime.t
Merge pull request #609 from rbt/master

Enable nanosecond times
02:42 sena_kun left 02:57 sena_kun joined
toddr AlexDaniel the test migration is done. I don't know who says go but if you guys are happy with that, I can send it to wherever and robert will setup the redirects. 03:30
03:57 evalable6 left, evalable6 joined 04:42 sena_kun left 04:55 sena_kun joined 05:55 sourceable6 left, coverable6 left, nativecallable6 left, quotable6 left, squashable6 left, notable6 left, bloatable6 left, committable6 left, releasable6 left, reportable6 left, benchable6 left, shareable6 left, greppable6 left, statisfiable6 left, unicodable6 left 05:56 bisectable6 left, nativecallable6 joined, squashable6 joined, sourceable6 joined 05:57 greppable6 joined, bisectable6 joined, notable6 joined, benchable6 joined, reportable6 joined, shareable6 joined, bloatable6 joined, statisfiable6 joined 05:58 coverable6 joined, releasable6 joined, committable6 joined 05:59 quotable6 joined, unicodable6 joined 06:41 sena_kun left 06:57 sena_kun joined 07:02 epony left 07:05 epony joined
lizmat Files=1295, Tests=109460, 209 wallclock secs (28.38 usr 8.04 sys + 2932.90 cusr 268.14 csys = 3237.46 CPU) 08:28
08:42 sena_kun left
Xliff Files=1295, Tests=109450, 179 wallclock secs (27.37 usr 3.54 sys + 3206.00 cusr 163.27 csys = 3400.18 CPU) # 2019.11-338-g8a0742eeb, 4 fails S11, S14, S24 and S32 08:51
lizmat: Am I on an earlier version? 08:52
lizmat my test run had one failure: the new DateTime test, which wasn't in my rakudo yet 08:53
and S32-str/utf8-c8.t failed, but that one always fails for me on MacOS 08:54
Xliff OK, cool. 08:55
lizmat twitter.com/_ahungry/status/1212945873333899264 08:56
maybe we should have a way to invite people to a specific repo only ?
08:56 sena_kun joined
Xliff Maybe github needs a way to bulk unwatch? ;) 08:57
lizmat perhaps, but that would require an action of the invitee... who didn't ask to be watching 200 repos 09:13
Xliff I meant: A mechanism for the invitee to bulk unwatch or bulk watch an orgs repositories. 09:14
lizmat well, the bulk watch was automatic by accepting the invitation
Xliff Yes. And there should be such a mechanism so that the invitee, who wants to be a member of the org, can not be swamped by content from said auto-watches. 09:15
Seriously, there needs to be stages from "I'll accept!" to "All in!" 09:16
nine Maybe there's a setting in the organization? 09:17
Geth rakudo: 56f30073e3 | (Elizabeth Mattijsen)++ | 4 files
Modernize extension of NativeCall.pm6
rakudo: 5c97532e34 | (Elizabeth Mattijsen)++ | 3 files
Modernize extension of NativeCall/Types.pm6
rakudo: 5a03783305 | (Elizabeth Mattijsen)++ | 3 files
Modernize extension of NativeCall/Compiler/GNU.pm6
rakudo: 8b223653e2 | (Elizabeth Mattijsen)++ | 3 files
Modernize extension of NativeCall/Compiler/MSVC.pm6
lizmat and that concludes modernizing all of lib
Geth rakudo: 7ce79f4d39 | (Elizabeth Mattijsen)++ | 3 files
Use .raku instead of .perl in lib/*.rakumod
rakudo: 5f335aa497 | (Elizabeth Mattijsen)++ | 13 files
Use .raku instead of .perl in all "make test" files
10:41 sena_kun left
AlexDaniel lizmat: oh well… 10:50
btw I also invited everyone to raku-community-modules 10:51
because I already had the script at hand and… well… yeah
I don't know if the best idea now :) 10:52
also, we have a small issue with permissions and stuff
I don't know if we should be strict about things like this, but in case of raku-community-modules the “base permissions” setting is set to “admin”
which is probably right if you want to manage webhooks and stuff for abandoned modules 10:53
but it's kinda weird that 300 people will have that
and it's similar for perl6/ org right now because each repo has different settings (and it's mostly random)
|Tux| Rakudo version 2019.11-373-g5f335aa49 - MoarVM version 2019.11-109-ge94893f89
csv-ip5xs0.748 - 0.767
csv-ip5xs-206.428 - 6.611
csv-parser37.962 - 38.935
csv-test-xs-200.418 - 0.426
test6.772 - 7.450
test-t1.766 - 1.857
test-t --race0.801 - 0.806
test-t-2029.939 - 30.060
test-t-20 --race9.120 - 9.320
nine So I'm gonna have to unsubscribe from all the repos I'm actually not interested in again? Somehow in hindsight it seems like just renaming the org would have been the better choice
AlexDaniel it used to be that creating repos in perl6/ automatically subscribed everyone to new repos but that's no longer the case I think 10:55
nine: what are you even talking about?
does it even apply to repo transfers?
does it actually work that way?
nine just infering 10:56
Could also be that everything is fine :)
AlexDaniel I think you have to be “owner” of the org to autosubscribe
but I'm not sure 10:57
10:57 sena_kun joined
AlexDaniel when problem-solving repo was created there were no watchers at all 10:57
besides maybe 7 people who are “owners”
now that I think about it, what is that tweet about? 10:58
where did they even find 200 repos?
AlexDaniel tests 11:01
I have just tested it by inviting a user to perl6/ org 11:03
I tried with both adding them to “perl6” team and without
there's no autowatch
nine cool!
AlexDaniel nine: but check github.com/settings/notifications 11:05
ok hold on, I think adding to a team actually does subscribe you 11:07
it doesn't subscribe to all 70 but it does subscribe to 50+ 11:08
because of that checkbox there, hmmmm
ooooooook 11:09
so it changed for perl6 because at some point I think I changed the default permissions to “read”
so now creating a new repo doesn't autosubscribe everyone to it, just the owners
but then adding perl6/ team *does* autosubscribe everyone 11:10
246 out of 333 people have that setting enabled and therefore are autosubscribed 11:11
lizmat: if they uncheck that setting in github.com/settings/notifications I think I can then kick that user out of any orgs they're in and reinvite, I think that'll save them some time trying to unwatch the repos 11:12
nine++ now I understand the situation better
lizmat: they can also “unwatch all” from here but that'd include any other repos too github.com/watching 11:14
nine AlexDaniel++ for dealing with all these boring organizational issues
AlexDaniel So TL;DR if you have that setting set in your own settings, you will always autowatch any repos we give you permissions to, and currently I don't see any way to fix that on our side besides no longer giving people permissions 11:16
the “200” repos in question is 50+ repos that people have access to in perl6 org, 5 repos in raku org and 70 repos in raku-community-modules 11:17
Xliff Ao I turn off "Automatically watch repositories" ? 11:24
AlexDaniel yeah if you don't want to watch all the repos you have access to
nine: also I think you're right, the moment we set permissions in the new org it'll autowatch unless you set the status to “ignore” or something like that 11:27
it seems like the autowatch feature triggers when you gain new permissions, which I think will be the case when we start transferring stuff 11:28
I can create a repo telling people to uncheck the setting 11:29
12:15 camelia left
Xliff Can someone restart camelia? 12:39
nine Method NQP::Config::configure_misc must be implemented by the language class at /home/camelia/rakudo-m-2/nqp/3rdparty/nqp-configure/lib/NQP/Config.pm line 563. 12:42
12:42 sena_kun left 12:48 camelia joined 12:57 sena_kun joined
lizmat can someone explain me (vrurg?) why this puns a Role: 12:59
m: role R { method m() { ... } }; dd R.foo
camelia Method 'm' must be implemented by R because it is required by roles: R.
in block <unit> at <tmp> line 1
lizmat and this doesn't:
m: role R { method m() { ... } }; dd R.gist
camelia "(R)"
lizmat is there a list somewhere of no-need-to-pun methods ? 13:00
nine lizmat: indeed, there is: github.com/rakudo/rakudo/blob/mast....nqp#L3910 13:03
lizmat nine++
Geth nqp: 7b1b0cebff | (Elizabeth Mattijsen)++ | 2 files
Give HLL/Actions|Grammar a "raku" method

And make the "perl" method a front for calling the "raku" method.
nqp: 530c7a6131 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MOAR to get the latest and greatest nine++ fixes
13:29 lucasb joined
Geth rakudo: 9a5579ff19 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Add "raku" as a method that will not cause punning

Also some comment changes
rakudo: 784daae2ca | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Add a "raku" method to BOOTSTRAPATTR
rakudo: 9ccfedb18a | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get all the latest goodies
sena_kun >Fetching [FAIL]: Text::Diff::Sift4:ver<1.4.0> from git://github.com/masterduke17/text-diff-sift4.git 14:16
is that just me or before zef accepted git:// urls just ok?
Geth roast: 6d23891a3a | (Elizabeth Mattijsen)++ | 9 files
Some s/perl/raku changes

These were necessary to make them pass on the s/perl/raku method switch soon to be done in Raku. Fortunately, these changes also pass on Raku
  *before* these changes.
rakudo: add2ec0d49 | (Elizabeth Mattijsen)++ | 50 files
Change all 'method perl' to 'method raku'

So the "raku" method is now leading, rather than a wrapper for the
  "perl" method. The "perl" method has now become a wrapper for the
  "raku" method.
sena_kun .seen MasterDuke17 14:29
tellable6 sena_kun, I saw MasterDuke17 2019-12-19T16:54:12Z in #raku-dev: <MasterDuke> oh, and handling return values from the libtommath functions, which warn if you don't, and we mostly weren't
Geth roast: aa4994a7f6 | (Elizabeth Mattijsen)++ | 222 files
Call method .raku instead of .perl in tests

Yes, this is a big commit. Some may still be missed :-(
lizmat *phew* Now on to the ecosystem :-) 14:34
AlexDaniel sena_kun: iirc it depended on your setup
sena_kun: also I think there was a way to write it in your meta6.json so that it always worked, maybe by using the https link? 14:35
14:42 sena_kun left
dogbert11 . 14:49
tellable6 2020-01-02T23:14:50Z #raku-dev <lizmat> dogbert11 S32-num/stress.t is a flapper for me
2020-01-02T23:16:56Z #raku-dev <lizmat> dogbert11 failure is typically 1.2873811240847445e-25 vs 1.2873811240847447e-25, aka the last shown digit
2020-01-02T23:18:23Z #raku-dev <lizmat> dogbert11 so it appears there *is* drift when roundtripping, and it is *not* load related
dogbert11 lizmat: do you think it's related to the recent 'val' changes? 14:50
lizmat I don't think so, but maybe we can bisect that
dogbert11 that is definitely worth a try 14:52
lizmat m: use Test; cmp-ok 3.782267030601322e-76.raku.Num.raku.Num.raku.Num.raku.Num, "===", 3.782267030601322e-76 14:54
camelia ok 1 -
lizmat hmmm
14:56 sena_kun joined
lizmat could be related to val() changes 14:57
and if that's true, it looks like I'm going to need a val() redesign :-( 14:58
grrr... 14:59
dogbert11: please make an issue for it :(
dogbert11 please don't shoot the messenger :) 15:00
vrurg run, dogbert11, run...
dogbert11 runs ... 15:01
lizmat nooooo :-) 15:04
dogbert11 R#3400 15:05
lizmat: github.com/rakudo/rakudo/issues/3400
lizmat dogbert11: github.com/rakudo/rakudo/blob/mast....pm6#L3084 15:06
am about to go away until later tonight, will look at it then
Geth problem-solving: c146268616 | (Aleks-Daniel Jakimenko-Aleksejev)++ | .github/ISSUE_TEMPLATE/issue-template-infrastructure.md
Remove second assignee from the template

Otherwise the template doesn't work at all. I think the syntax was wrong, but this fix will do for now.
problem-solving: 834840e43a | (Aleks-Daniel Jakimenko-Aleksejev)++ | 2 files
Rakufy the templates
problem-solving: 48d036cb42 | (Aleks-Daniel Jakimenko-Aleksejev)++ | README.md
Rakufy the README
AlexDaniel but why is it flapping?? O_o 15:43
Geth ¦ problem-solving: melezhik assigned to rba Issue Test Raku modules against different OS, Rakudo versions github.com/perl6/problem-solving/issues/144 15:47
bartolin vrurg: I think I have a fix for G#3399. I'm not really sure why, but removing the 'return' here helps: github.com/rakudo/rakudo/blob/add2...u.pm6#L826 Running a spectest now. 15:49
maybe it happens because return does a nqp::p6recont_ro (github.com/rakudo/rakudo/blob/add2....pm6#L37). And MoarVM uses the code from github.com/rakudo/rakudo/blob/add2...ns.nqp#L44 that doesn't have a return. 15:52
vrurg bartolin: looks reasonable. 16:02
vrurg doesn't remember if the return there is mine...
bartolin yeah, it's new there. I'll commit the fix in a moment. 16:03
vrurg git confirms. Wonder why I put it there because usually I avoid it. 16:05
bartolin: great job, thanks!
vrurg is afk
Geth rakudo: 262b6fc7ae | (Christian Bartolomäus)++ | src/core.c/Mu.pm6
Don't use explicit return in dispatch:<::>

Fixes #3399 (for the JVM backend). Please note that MoarVM uses different code for this purpose:
sena_kun AlexDaniel: 1)how fast bisectable builds for new commits are done? 2)where can I check list of all builds? 16:30
AlexDaniel sena_kun: there's a build for every rakudo commit on master branch since 2014.01 16:32
sena_kun: new builds appear within a few minutes after they're committed to master
e: say 42
evalable6 42
AlexDaniel e.g. that last commit is already there
sena_kun: last time I checked there were over 9000 builds 16:33
c: 2014.02 say 42
committable6 AlexDaniel, ¦2014.02: «42␤»
AlexDaniel c: 2014.02^ say 42
committable6 AlexDaniel, ¦2014.02^: «42␤»
AlexDaniel yeah
sena_kun AlexDaniel: great! That's fast. Great. I am playing with Blin right now and got "no end point build available" for a couple of runs, hence the question.
AlexDaniel sena_kun: iiinteresting. What did you give it for --new ? 16:34
sena_kun AlexDaniel: nothing. I mean, it took latest from master, but the build wasn't ready for latest commit yet. Now it is available and it works now, hence no issue.
AlexDaniel ahhh 16:35
sena_kun: right… well, two things can improve it: support for webhooks and on-demand builds 16:36
these things were on my to-do list since forever 16:37
goddamn software is never ready :)
well, I mean, complete
sena_kun AlexDaniel: it is very cool already, I think. :) 16:38
though I can't reproduce "already installed" bug locally and it kind of sucks, since I wanted to look into that, that's annoying...
AlexDaniel evalable6 doesn't advance to the next build until it's ready, but of course if you're running blin locally then it kinda doesn't know
sena_kun: isn't purging the installed/ dir enough to fix that?
Geth roast: 93ada5fa3b | (Christian Bartolomäus)++ | 2 files
[JVM] Fudge some newly added test
16:43 sena_kun left
Geth Blin/enchance: 4 commits pushed by Altai-man++ 16:53
Blin: Altai-man++ created pull request #16:
Introduce some goodies for making the release checking process easier
16:57 sena_kun joined
AlexDaniel sena_kun: review complete :) 17:00
sena_kun bisectable6: say "ASCII text" ~~ any("ASCII C program text", "C source, ASCII text") 17:12
bisectable6 sena_kun, On both starting points (old=2015.12 new=262b6fc) the exit code is 0 and the output is identical as well
sena_kun, Output on both points: «False␤»
sena_kun hmmmm
Geth Blin/enchance: 7c49129173 | Altai-man++ | bin/blin.p6
Remove accidental ws character
Blin/enchance: 9f3a324de9 | Altai-man++ | bin/blin.p6
Fix condition
AlexDaniel sena_kun: but now $_.any doesn't do the thing! 17:20
sena_kun: I'm actually not sure what this piece of code is about, hmmm… and also why is it being changed
it used to work, I think? Now it's giving warnings about Nils or something?
sena_kun AlexDaniel: changed because it gives warnings.
AlexDaniel ah it's if you give it a list of modules 17:21
sena_kun I don't know if it used to work, I think it did, but it gives warnings for me every time and this patch makes this awooo go away.
AlexDaniel sena_kun: also I don't think it's possible to split the condition into two lines 17:23
the logic here is to continue to mark the module as needed if it was specified on the command line, right? 17:24
sena_kun AlexDaniel: de Morgan laws are no more?
AlexDaniel sena_kun: hehe :) not here 17:25
sena_kun: because if it fails the condition then it'll do “next”
if you can rewrite it in a better way I'll appreciate it
sena_kun AlexDaniel: I'll look at it some more, maybe I am not fresh enough after a short vacation. :S 17:26
AlexDaniel that whole piece of code is to mark modules that were given on the command line
sena_kun: I had to stare at that unreadable shit for about five minutes actually :D
without understanding anything during that whole time
why is there an .any on a single string I also have no idea 17:27
sena_kun: it's inverted inverted logic, and the fact that junctions by themselves are also inverted doesn't really help 17:28
but it's simple. If it's in @specified-modules, then mark it. If it's a $custom-script, then mark it too
I guess that's exactly how it should be written :D 17:29
sena_kun let's see...
AlexDaniel if .key eq @specified-modules.any or .key eq $custom-script { .value.tail.needify }
sena_kun I am trying to make perl6-File-LibMagic pass its tests too, but there is some oddness happening.
AlexDaniel $custom-script was added later as an afterthought which is why the code turned into this crazy mess 17:31
in that section
Geth Blin/enchance: bc0ee97898 | Altai-man++ | bin/blin.p6
Simplify condition
AlexDaniel sena_kun: so I wonder 17:46
sena_kun: Movieinfo is listed in that PR
sena_kun: but it's not here: github.com/perl6/ecosystem-unbitro...+is%3Aopen
sena_kun: ===> Install [OK] for Movieinfo:ver<0.1.2> 17:47
why are we ignoring a module that used to install OK?
sena_kun AlexDaniel: are you installing it locally or using `zef install Movieinfo?
AlexDaniel sena_kun: that's from the last blin run
===> Install [OK] for Bencode:ver<0.1.1>
sena_kun >"===> Searching for: Movieinfo\nEnabled fetching backends [git path curl] don't understand git://github.com/palapitta/Movieinfo
AlexDaniel Enabled fetching backends [pswebrequest path curl wget] don't understand git://github.com/palapitta/Movieinfo 17:48
You may need to configure one of the following backends, or install its underlying software - [git]
that's just zef acting wrong, no? 17:49
I don't know which version of zef I have but it used to work in the past
and then, even if it's not supposed to work, there's no PR
17:52 Altai-man_ joined, sena_kun left
Altai-man_ AlexDaniel, there is a PR dated back to 2016 adding the mandatory field ignored, seems abandoned to me. 17:52
tellable6 2019-12-28T01:43:36Z #raku-dev <vrurg> Altai-man I made #3311 a blocker. Even though it's macos specific but it's a cornerstone feature which is broken.
AlexDaniel Altai-man_: maybe, maybe not, that field makes no difference whatsoever anyway 17:54
I would've shrugged on that PR as well and moved on with my life
Altai-man_ AlexDaniel, there is also exactly 0 commits from the author last year.
AlexDaniel well that yeah
Altai-man_ and in 2019
and in 2018 17:55
...and in 2017
AlexDaniel but let's no go this route of “we submitted a PR and it wasn't merged so we ignore the broken module”
because the module is not broken
it doesn't need that useless field, and the link used to work fine
Altai-man_ AlexDaniel, it is not installable by Blin --> it is a false negative.
AlexDaniel Altai-man_: it was installable, that's my point 17:56
Altai-man_ AlexDaniel, well, I already was wondering why zef doesn't know about git:// links.
and then, hmm...
maybe we used panda in 2016?
and panda worked with git:// links?
AlexDaniel I repeat, it used to install OK in Blin 17:57
just a few months ago
Altai-man_ AlexDaniel, a zef ticket?
AlexDaniel I disagree with the whole idea of starting to ignore modules because we changed our tooling and then claimed the modules broken yet never even submitted a ticket or a PR
old modules are great for tracking changes in rakudo and spotting regressions, they have a lot of crazy code which works wonderful as tests :) 17:58
Altai-man_ AlexDaniel, do you assume that if a module is broken for last three years, one day it'll somehow pass the test? that itself would be a regression... which kind of does proves your point, admittedly. 17:59
AlexDaniel Altai-man_: what do you mean last three years 18:00
I repeat for the nth time, it installed OK like three months ago in Blin
both Bencode and Movieinfo did
there are the first two I looked at, I guess there are more modules like this in the list
the mandatory field makes no difference I hope
and I don't even know why it's mandatory 18:01
the PRs adding `6.*` are just useless, in my opinion
like, “you *have* to write this because reasons”
Altai-man_ hmmm 18:02
AlexDaniel but I'm pretty sure it doesn't affect it anyway, it just can't read the git link
Altai-man_ let's approach it from the other side
AlexDaniel which is probably something that has to be fixed in META6.json, but *we* broke it in our tooling, the module was fine for all these yeasr
Altai-man_ my side is that last month I had to manually check 400+ failing modules, I guess that's the amount we did in last squashathon, maybe more. I'd like not to repeat that. There are a couple of things I know. First one is "module is already installed". Can't reproduce it locally. Second one is modules which are simply abandoned, like the one you mentioned. 18:04
the second option is to investigate why are you running Blin and having 170 modules, I am running it on the machine I have access to and getting 400. 18:05
AlexDaniel I'm not getting 170 modules, that was like three months ago 18:12
if some change in zef did something then I guess I'd also get 400 today
Altai-man_ there were no changes in zef related to this
AlexDaniel ok but somehow it stopped working, right? 18:13
Altai-man_ ugexe claims p6c.org module list used to mangle urls to append `.git` at the end (which is missing and that's why it is broken). but we don't have p6c.org for longer than two months.
AlexDaniel Altai-man_: here's the file I'm looking at github.com/perl6/problem-solving/f...a.json.txt 18:14
there's no date in it, which is my failure of course, but it's from 2019
Altai-man_ AlexDaniel, interesting... 18:15
I also see 24 "already installed" entries...
AlexDaniel Altai-man_: but does it matter? 18:16
18:16 patrickb joined
Altai-man_ AlexDaniel, it's not, just a side note... 18:16
AlexDaniel, I see what's wrong... probably 18:19
let me check one more thing...
Geth rakudo: e5627b83a3 | (Christian Bartolomäus)++ | .travis.yml
Skip 'make test' on JVM backend

It doesn't finish in a reasonable time and there are various failures for those tests that were run. It's just a big waste of resources.
  *If* running 'make test' on the JVM backend gets in a better shape in
the future, it could be re-enabled easily.
bartolin hopes the above commit works as expected and doesn't change the travis runs for moarvm
Altai-man_ AlexDaniel, I see that in blin.p6 @sources has `ecosystem-api.p6c.org/projects.json`, which is AFAIK dead. And if you look at `Movieinfo` module, the URL is indeed with `.git` added. the alternative zef uses is raw.githubusercontent.com/ugexe/Pe...p6c1.json. And data there is not mangled. 18:21
AlexDaniel hmmm
[Coke] (jvm make test) should we fudge it so that we are at least running the passing tests? 18:22
Altai-man_ after spring (was it spring?) ecosystem server dying, was ecosystem data update code ever brought back to life?
AlexDaniel Altai-man_: no idea
rba: ↑?
bartolin [Coke]: I don't think, it's worth the trouble. One problem is the broken EvalServer (RT #132104). It takes really, really long to run the tests and I doubt we'll be fast enough for the travis timeout. One idea would be to run only a subset of the tests. 18:25
AlexDaniel bartolin: nice that you're here!
bartolin: can you review github.com/Raku/p6rt ?
just take a look at the tickets and tell if it looks alright 18:26
bartolin [Coke]: But then I'm nearly sure that nobody has looked at the travis tests for the JVM backend for a very long time.
AlexDaniel: yeah, I can do that (probably a bit later today) 18:27
fine, at least I didn't mess up the tests for moar: travis-ci.org/rakudo/rakudo/jobs/632392114 18:29
[Coke] bartolin: I just happened to build a copy of jvm today because I had a thought about targetting a 4th backend. 18:30
bartolin [Coke]: Oh, interesting! I'm still trying to fix regressions if I see them, but I'm only looking at the nqp tests and the spectests. No longer running spectests daily, but from time to time. 18:32
[Coke] I was able to do a build and 'make install' locally, but trying to run it interactively and run 'say 3+4' in the REPL basically hung. 18:51
(the REPL did launch, though)
bartolin oh, I remember seeing that, too. I suspected it to be a bug with buffering of stdin. If you type ^D after executing your command you should get an answer (the REPL exits, though) 18:53
I'm mostly using it with -e. That works fine. 18:54
hah, I've even made a note locally "REPL: doesn't see NL: > say 42; hangs" ]: 18:56
[Coke] (y)
Xliff What does this error mean? "Cannot find method 'package_at_key' on 'BOOTHash': no method cache and no .^find_method" 19:13
rba Altai-man_: There is a server under my controls servering ecosystems-api.p6c.org. Yet not sure if this update mechanism is still working fine. 19:16
tellable6 2020-01-02T18:07:45Z #raku-dev <patrickb> rba I have created binary builds for Linux, MacOS and Windows for the rakudo 2019.11 release. They can be downloaded here: patszim.volans.uberspace.de/patclo...EJLQDZcjbf
2020-01-02T18:08:43Z #raku-dev <patrickb> rba You can put them on stage-rakudo.rakulang.site/
2020-01-02T18:09:14Z #raku-dev <patrickb> rba That's the last missing bit that was standing in the way of the new rakudo.org website as far as I know.
2020-01-02T18:10:17Z #raku-dev <patrickb> rba Some testing of those builds would be welcome of course.
2020-01-02T22:26:20Z #raku-dev <patrickb> rba The stage-rakudo.rakulang.site/dl -> github redirect seems to be active again. Can you disable that (for the staging website only) again?
rba Altai-man_: Think I need some help with this. 19:17
bartolin toddr AlexDaniel: if I'm not mistaken, the migration script doesn't re-open a GH issue when the status on RT was changed from 'resolved' to 'open'. For example rt.perl.org/Ticket/Display.html?id=72820 is still open, while github.com/Raku/p6rt/issues/1498 is closed
though the status change is there: github.com/Raku/p6rt/issues/1498#i...-570363850 19:18
toddr looks like it's getting confused because it was closed then opened later 19:19
bartolin looks further -- just wanted to note this one early 19:20
toddr LOL so the export lists a date in the resolved field even though status=open 19:21
bartolin well, the ticket *was* resolved on 2016-07-06 (but only for 15 minutes) 19:23
rba Altai-man_: Current zef is not using the ecosystems-api url any more. 19:24
19:38 timotimo joined
toddr bartolin thanks for the catch. you can see the fix in github.com/Raku/p6rt/issues/6695 and github.com/Raku/p6rt/issues/6694 19:41
bartolin toddr: another question: it looks like recent changes to tickets aren't taken into account. I guess the export was run on a database export from rt. Do you know the timestamp of that export? Example: rt.perl.org/Ticket/Display.html?id=76930 was closed on 2019-08-29, github.com/Raku/p6rt/issues/2003 is still open
toddr AFAIK that queue was locked in June 19:42
bartolin toddr++ (fast fix)
toddr I can ask for a new export.
bartolin oh, June already?! That's strange, because the update to rt.perl.org/Ticket/Display.html?id=76930 shouldn't have been possible, should it? 19:43
toddr: I think that would be good. I closed a couple of issues (in the order of 200) in the end of August. 19:44
(I was assuming that those changes would make it into the new 'old-issue-tracker') 19:45
toddr :\
I would REALLY prefer to not get a new export. I had to fix quite a bit of mixed encoding problems in the export files. 19:46
can you just re-close them after the switch to github?
bartolin oh, I see. Yeah, that is totally possible.
toddr No idea who's in charge so I'm still waiting for a good to go. Possibly others will find problems with the test migration? 19:48
bartolin if you could get an exact date of your export, I'll take care of the changes that where made after said date. 19:49
AlexDaniel asked me to take a look at github.com/Raku/p6rt earlier today. things look very promising so far. I hope I won't find anything problematic in the next few hours :) 19:51
19:52 sena_kun joined 19:53 Altai-man_ left
Xliff When you re-export a package by using "use Package; sub EXPORT { %(PACKAGE::EXPORT::DEFAULT::) }", it works, but doesn't have a package itself so it can be re-exported. How can I fix that? 19:54
See this example: github.com/Xliff/p6-GLib/blob/mast.../Types.pm6
I can "use GLib::Raw::Types" and it will DWIM, but I cannot re-export the symbols from GLib::Raw::Types using the same mechanism, because GLib::Raw::Types isn't a package.
So what am I missing? 19:55
moritz cross-posted from #raku, FWIW
AlexDaniel toddr: I left a comment here wrt the final destination and stuff github.com/perl6/problem-solving/i...-570652148 20:03
toddr: let's give it maybe a day for people to take a look and then you're free to press the big button :) 20:05
toddr thanks
The labels will be a little bit of a mess. 20:06
I'd need to probably make an additional mapping to make it work.
I'll look at it a little later. 20:07
AlexDaniel we can tweak them later, but it's annoying that github will keep showing the old colors and stuff in history
toddr oh? I thought it updated if you altered the label
AlexDaniel like the action of labeling the ticket in comments will have the old color
toddr you sure that's not just a cache issue?
AlexDaniel let's see if I can find an example… :)
bartolin toddr: would it be easy to look at the 'testneeded' tag in RT and make that a label? (I think there are a couple of RT tickets tagged with 'testneeded'. I'm looking for an example.) 20:08
AlexDaniel oh yeah that's an important one
bartolin example for testneeded: rt.perl.org/Ticket/Display.html?id=133762 and github.com/Raku/p6rt/issues/6689, respectively 20:10
toddr It sounds like you want me to merge in these labels gist.github.com/toddr/160b0c4ee144...5819e61474
or do I strip out Bug? 20:11
AlexDaniel toddr: you're right, it's no longer an issue! It used to be that messages like “AlexDaniel added the `tests needed` label” always had the old color and old name, but it seems to be correct now
I'm pretty sure it wasn't just cache 20:12
toddr: `bug` wasn't really used consistently, so I think it's safe to throw it away
bartolin (lables) skimming through that list I'd say only testneeded is useful 20:13
AlexDaniel if we care about old tickets then maybe some others too
but personally I don't
like, there were many tickets for patches in the past 20:14
well, maybe not many, but some :)
bartolin well, if it's easy to import the tags, it would make things more consistent to import all of them. 20:15
toddr Including or excluding "Bug" 20:19
bartolin I *think* most tickets tagged with 'Bug' will also have [Bug] in the subject. If that duplication doesn't cause problems, I'd vote for including. 20:20
bbl &
AlexDaniel heh I think ‘Bug’ is just noise but I don't want to make your life harder :D 20:21
including `bug` means pretty much every ticket will have it… 20:22
bug bug bug bug bug bug bug
125138.json: "Tag" : "Bug\nBug", 20:23
toddr: wait, but how do you create tickets/comments with timestamps in the past? 20:25
I didn't know this was possible 20:26
20:45 patrickz joined 20:46 patrickb left
bartolin toddr AlexDaniel: I've looked at the test migration (comparing on a sample basis) and have some (minor) remarks. All in all the result looks very good to me and I'd be okay with pressing the big button. But maybe other folks will take a look as well. (more) 20:55
toddr I have time stamps in the past by using a super secret API I'm not allowed to tell you about. 20:57
bartolin (1) Sometimes URLs are not clickable on GH -- sometimes they are. I could look up an example, but I don't think this is worth any trouble
toddr where are they not clickable? 20:58
bartolin (2) Sometimes non-ASCII characters are garbled -- mostly they are ok. One example: rt.perl.org/Ticket/Display.html?id=132452 vs. github.com/Raku/p6