dalek ast: 907379d | usev6++ | S (2 files):
Add fudging for rakudo-j for new tests
05:13
AlexDaniel I hope nobody needed bisectable while it was down :) 06:14
there are many cool new features now
bisect: say (^∞).grep({ last })[5]
bisectable AlexDaniel: Exit code is 0 on both starting points, bisecting by using the output
AlexDaniel: bisect log: gist.github.com/2d8b6006eaf4781865...79ef765ae3
AlexDaniel: (2016-03-18) github.com/rakudo/rakudo/commit/6d120ca
AlexDaniel one is that it provides the log 06:15
with *all* information
bisect: say ‘hello world’
bisectable AlexDaniel: On both starting points the exit code is 0 and the output is identical as well
AlexDaniel: Output on both points: hello world
AlexDaniel basically, it is much more transparent now
and if the output is too long it will gist it 06:16
bisect: .say for ^200
bisectable AlexDaniel: On both starting points the exit code is 0 and the output is identical as well
AlexDaniel: gist.github.com/b7db9754fc51f4f7c6...7fa18a6205
AlexDaniel bisect: good=2016.03 bad 2016.02 say (^∞).grep({ last })[5] # swapped good and bad revisions
bisectable AlexDaniel: Exit code is 0 on both starting points, bisecting by using the output
AlexDaniel: bisect log: gist.github.com/256227743bd646f2d8...da2789c33e
AlexDaniel: bisect init failure
AlexDaniel ↑ this error message comes from git. Now you can see it! 06:17
but that's not all… :) Ever wanted to m: some code but the output was a bit too long?
committable: HEAD say rand for ^100
committable AlexDaniel: gist.github.com/0d89cfa80a84bf4e72...4f4dd7ffda
AlexDaniel one cool thing about committable is that it allows multiple revisions to be specified 06:18
committable: 2015.10,2016.02,HEAD say $*VM
committable AlexDaniel: |«2015.10»: moar (2015.10)␤|«2016.02»: moar (2016.2)␤|«HEAD»: moar (2016.06.9.g.8.fc.21.d.5)
AlexDaniel and even though it does not run anything in parallel, you can already go wild 06:19
commit: HEAD~250..HEAD say round(rand*3)
committable AlexDaniel: Too many commits (363) in range, you're only allowed 300
AlexDaniel ah yeah
commit: HEAD~200..HEAD say round(rand*3)
not too wild
but still
slooooooooow because not parallel, but already useful
committable AlexDaniel: gist.github.com/48a9797ffaba774b0f...93d058e964 06:20
AlexDaniel note that it groups commits by the output
anyway, that's it! Have a nice day everyone
MasterDuke++ for 50% of these changes
masak AlexDaniel++ # improvements 06:34
though don't you think "commitable" is a confusing name, since it doesn't actually allow committing? ;) 06:35
er, "committable"
AlexDaniel masak: how else should we call it? Executable? :D
masak "mucoevbo" ("multi-commit eval bot") ;) 06:38
or just "multicommit"
hm, I guess the "multi" ain't the main feature, though 06:39
how about calling it "onbot"? :D it evals something "on" a certain commit
AlexDaniel we will see what is the main feature, hard to tell right now
anyway, “committable” is how MasterDuke contributed it, and I don't mind… but pull request are welcome, right? ;) 06:40
Zoffix .ask psch is JVM still seriously busted or should it be fully tested as part of this weeks release?
AlexDaniel committable: source
committable AlexDaniel: github.com/perl6/bisectbot
Zoffix .tell jnthn just a reminder that tomorrow is a Rakudo release and we'll need a MoarVM release cut. Thanks. 06:41
ugh
Never a bot around when you need one!
masak bots are so lazy 06:42
Zoffix .ask psch is JVM still seriously busted or should it be fully tested as part of this weeks release?
yoleaux2 Zoffix: I'll pass your message to psch.
Zoffix .tell jnthn just a reminder that tomorrow is a Rakudo release and we'll need a MoarVM release cut. Thanks. 06:43
yoleaux2 Zoffix: I'll pass your message to jnthn.
masak .tell yoleaux2 that circularity is problematic, m'kay? 06:44
yoleaux2 masak: Thanks for the message.
Zoffix hah 06:45
.ask nine have you had any luck with DESTDIR ticket rt.perl.org/Ticket/Display.html?id=128548 ? I recall you mentioned it's a release blocker. How should we proceed here?
yoleaux2 Zoffix: I'll pass your message to nine.
masak .tell yoleaux2 to thank me for the message 06:56
yoleaux2 masak: Thanks for the message.
Zoffix m: say so "07\x[308]2" ~~ /^ <[0..9]>+ $/ 07:07
camelia rakudo-moar c9dcc9: OUTPUT«True␤»
Zoffix This is a bit scary. 07:08
m: "\x[308]".uninames.say
camelia rakudo-moar c9dcc9: OUTPUT«(COMBINING DIAERESIS)␤»
Zoffix m: say so "07\x[308]2" ~~ /^ <[\x[30]..\x[39]]>+ $/ 07:12
camelia rakudo-moar c9dcc9: OUTPUT«True␤»
Zoffix o.O
m: Date.new("२०१६-०७-१६").say 07:14
camelia rakudo-moar c9dcc9: OUTPUT«2016-07-16␤»
nine Zoffix: please don't do a release until I fixed the regression. I think I'm almost done and should have quite some time for rakudo today. If I don't make it, I will revert the cause of the regression, but you never know when real life interferes... 07:15
yoleaux2 06:45Z <Zoffix> nine: have you had any luck with DESTDIR ticket rt.perl.org/Ticket/Display.html?id=128548 ? I recall you mentioned it's a release blocker. How should we proceed here?
nine Hah! Just answered :)
Zoffix nine, no problem
AlexDaniel bisect: good 2015.10 my $p1 = start die "I’m dead"; my $p2 = start ~"I’m alive"; try await Promise.allof($p1, $p2); say .result for grep { .status ~~ ‘Kept’}, $p1, $p2; 07:23
bisectable AlexDaniel: On both starting points the exit code is 0 and the output is identical as well
AlexDaniel: Output on both points: Iâm alive
AlexDaniel great, encoding issues
dalek kudo/nom: 2c69ab7 | (Zoffix Znet)++ | docs/release_guide.pod:
Update leap seconds instructions in the Release Guide
07:56
kudo/nom: 325b43f | lizmat++ | docs/ChangeLog:
Mention improvements of last night
07:58
kudo/nom: da0973d | lizmat++ | src/core/Array.pm:
Remove Array.cow for now

Nobody's really said anything about it, and there are no tests for it yet. So give us the option of not making it a feature by removing it for the 2016.07 release.
08:01
lizmat afk for the rest of the day& 08:02
Zoffix bisect: "Foo".comb.count-only 08:32
bisectable Zoffix: On both starting points the exit code is 1 and the output is identical as well
Zoffix: Output on both points: Method 'count-only' not found for invocant of class 'Seq'␤ in block <unit> at /tmp/7QA9mmmvbo line 1␤
Zoffix bisect: good=053729c bad=HEAD "Foo".comb.count-only 08:33
bisectable Zoffix: On both starting points the exit code is 1 and the output is identical as well
Zoffix: Output on both points: Method 'count-only' not found for invocant of class 'Seq'␤ in block <unit> at /tmp/nVuFY3OyfF line 1␤
timotimo count-only is in the iterator, not in the seq 08:45
psch . 09:11
yoleaux2 06:43Z <Zoffix> psch: is JVM still seriously busted or should it be fully tested as part of this weeks release?
psch Zoffix: afaict we install and kinda sorta get through spectest. bartolin++ did lots of fudging there recently 09:12
Zoffix: 'make test' is mostly fine, except NC is broken in part
Zoffix Are usev6 and bartolin the same person?
psch Zoffix: and the interop changes (seperating out rt.jar classes vs "classes loaded from somewhere not in the classpath" aren't quite done
yeah
i'm kinda confident i can fix the latter today, if not tomorrow 09:13
that's kind of my for-relaxation thing currently, NC probably after that, although that might well take a week or so and probably shouldn't block the release 09:14
in general i'd advice not to release r-j as of now
Zoffix cool. thanks 09:15
psch i mean, we probably could, but my impression (without looking closely) is that the amount of fudges and skips are (a) quite significant atm and (b) affecting important areas 09:18
like, i think we still have some problem with CX::Return in some spots
and other problems with implicit return of specific things... 09:19
stmuk wonders if sexy-coder-robot is all fired up for the release 10:10
Zoffix Sure :) 10:16
dalek kudo/nom: 51a6118 | (Zoffix Znet)++ | docs/ChangeLog:
Populate ChangeLog

Logs changes from commits: b7c8528 and bd0fd78
10:55
psch nine: i need a hint. why does install-core-dist.pl want to create .precomp below a .jar? 11:10
nine: reason is, i apparently don't have a choice but to include the jars (i.e. the asm*.jar, nqp-runtime.jar, etc) in nqp::jvmclasspaths, because otherwise i lose them, which breaks interop for JavaRuntime 11:11
nine: but i don't get why we need .precomp there, nor where and why that happens
nine psch: CompUnit::PrecompilationStore::File currently creates the precomp directory in its constructor. 11:23
psch: If your Java repository creates a CompUnit::PrecompilationRepository with the jar's path as $.prefix, this will be passed to the precompilation store 11:24
psch nine: the actual issue seems to be that the RepositoryRegistery creates a PrecompStore::File for each entry in the CLASSPATH 11:26
nine: and that's where we have e.g. asm-4.1.jar 11:27
nine psch: ack PrecompilationStore::File src
psch: looks to me like the only places where we create a PrecompilationStore::File are the precomp-store methods of repositories? 11:28
psch nine: github.com/rakudo/rakudo/blob/nom/...try.pm#L66 is what i'm looking at 11:30
nine: if i change nqp::jvmclasspaths to not include the *jar paths, i don't get the .precomp problem 11:31
s/include/emit/
nine psch: I guess that creates CompUnit::Repository::FileSystem instances for those class paths. That's clearly no longer a good idea. 12:46
psch: either just remove it, or change it so it will create Java repositories instead. 12:47
psch: in the latter case just prefix the path by 'java#' or 'jar#' or whatever you use as short-id
psch yeah, i think the latter could work 12:48
i'm a bit concerned that it'll do the whole interop shtick then, though, which we don't really want for e.g. asm-4.1.jar
but i don't really see through the interaction of the jvms classpath and the hll-level lib paths :| 12:49
[Coke] bestpractical.com/blog/2016/7/tagging-in-rt - a real replacement for us abusing [BUG] in the subject. 13:18
Frameless Creepy you mentioning it just as I'm writing a blog about tickets and and section on RT tagging in particular. 13:19
Quit spying on me! :)
[Coke]: BTW, did we get the [@LARRY] or similar tag on RT? Would be nice to mention it in my post, if we got it 13:20
"...And once the values are added, they become links to a search of all active tickets which that tag...." 13:22
That would be great. Links to get tickets with a particular tag.
[Coke] Frameless: just need to agree what the tag is. 13:24
Frameless: we can probably get that link now by doing a subject search. (checking) 13:25
Frameless It'll likely be the one that shows CSRF vuln warning 13:26
[Coke] of course. but here: rt.perl.org/Search/Results.html?Fo...me__%27%2C 13:27
%0A%27__Priority__%27%2C%0A%27__NEWLINE__%27%2C%0A%27%27%2C%0A%27%3Csmall%3E__Requestors__%3C%2Fsmall%3E%27%2C%0A%27%3Csmall%3E__CreatedRelative__%3C%2Fsmall%3E%27%2C%0A%27%3Csmall%3E__ToldRelative__%3C%2Fsmall%3E%27%2C%0A%27%3Csmall%3E__LastUpdatedRelative__%3C%2Fsmall%3E%27%2C%0A%27%3Csmall%3E__TimeLeft__%3C%2Fsmall%3E%27&Order=ASC%7CASC%7CASC%7CASC&OrderBy=id%7C%7C%7C&Query=Subject%20LIKE%20%27%5BPERF
%5D%27%20AND%20Queue%20%3D%20%27perl6%27%20AND%20(%20%20Status%20%3D%20%27new%27%20OR%20Status%20%3D%20%27open%27%20)&RowsPerPage=0&SavedChartSearchId=new&SavedSearchId=new
oof. one sec.
- goo.gl/AvGhwD [rt.perl.org]
(CSRF) I'll ask the team at rt.perl.org to see if we can whitelist rakudo.org and *.perl6.org 13:29
perlpilot [Coke]: I think if you choose [@LARRY] and start using it, it will be agreed :) 13:30
Frameless Yeah :)
[Coke] Frameless: perfect, sold, use it. 13:31
Frameless Cool. 13:32
dalek kudo/nom: e721ec7 | coke++ | README.md:
add new [@LARRY] tag
13:33
Frameless [Coke]: so we'd just put that in the subject line? 'cause there's also the RT tags thing that lets you pick from a predefined list of tags. 13:34
[Coke] I don't think those tags are installed on rt.perl.org yet, are they? 13:37
oh, that predefined tag list.
that's different.
updating that list requires a ticket with the RT admins, so I tend to avoid it.
(and the original list came directly from perl5's list, so not sure how much of them are used/relevant) 13:38
Frameless Ah. Sure. The in-subject tags are preferred by me too, because they are the only thing my RT::REST::Client understands :)
[Coke] nine: is rt.perl.org/Ticket/Display.html?id=128371 resolved? it says it's another copy of a resolved precomp bug. 13:48
Frameless And there you go, the first @LARRY tag: rt.perl.org/Ticket/Display.html?id=128542 13:49
[Coke] is pmichaud the only one able to address 128423 ?
perlpilot guesses so 13:50
[Coke] ah, crap. @LARRY is the first tag with not just [A-Z] 13:51
Frameless Change it to [LARRY]? 13:52
[Coke] eh, it's fine.
RT: 1322; @LARRY: 1; CONC: 7; GLR: 6; JVM: 70; LHF: 1; LTA: 79; NEW: 876; NYI: 28; OSX: 6; PERF: 16; POD: 9; PRECOMP: 7; RFC: 22; SEGV: 24; STAR: 1; TESTNEEDED: 10; TODO: 8; UNI: 10; UNTAGGED: 655; WEIRD: 3
Frameless OK :)
[Coke] liz just merged in "is mangled" for nativecall; do we need to bikeshed the name or worry about 6.c compatibility with this one? 13:54
nine [Coke]: the testcase for 128371 is incomplete, so I cannot run it. But I strongly suspect that it's indeed fixed, too. 14:07
[Coke] "how is zoffix replying to my ticket to the perlbug admins??" "oh, because that's not where that ticket went." 14:09
nine: Can you reply to the user and ask them to either test with HEAD and/or ask them for whatever bits are missing to allow us to run the test? 14:10
(reply, cc' list, etc.) 14:12
nine [Coke]: done
dalek kudo/nom: 78ba3df | (Zoffix Znet)++ | docs/ChangeLog:
Populate ChangeLog

List changes introduced by Dependency Version 2
14:13
[Coke] .tell TimToady - we've started tagging RTs that probably need design review with [@LARRY]; one more thing for you to look at if you get some time. other @larrys will pitch in as well. (many more tags to come)
yoleaux2 [Coke]: I'll pass your message to TimToady.
Frameless New blog post: "[Part 2] A Date With The Bug Queue or Let Me Help You Help Me Help You": perl6.party/post/A-Date-With-The-Bu...ou--Part-2 17:02
AlexDaniel “This isn't the interface for the latest NASA space probe.” – Haha, so true. That being said, I kinda like how advanced it is. The only problem is that I don't need 90% of these features most of the time, so I wish the “simple search” thingy was slightly more advanced. 17:06
Frameless++ 17:10
bartolin Frameless++ # work on RT queue and blog posts 17:12
bartolin liked Part 1 and will read Part 2 later
Frameless: btw, yes, I'm usev6 on github
Frameless bartolin: is my adding tests annoying because they fail on JVM and I don't test the JVM and you have to fudge them? :) 17:13
bartolin Frameless: no, it's fine. I'm happy to see new tests :-) 17:14
Frameless k :)
bartolin I have one question, though: did you see that we have t/02-rakudo/repl.t in the rakudo repo? 17:15
maybe it's better to move tests for the REPL out of roast (since it does not belong to the language specification) 17:16
or does it?
[Coke] if it's in 6.c, it's -in- the language spec at the moment. 17:17
but yes, moving those tests to rakudo-actual is not a bad idea, depending on the tests.
in some cases, it's "this used to fail in the REPL" vs. "this thing that should only ever happen in the REPL works 17:18
bartolin [Coke]: in 6.c we only have one basic sanity test
[Coke] well, it's not automatically going in to 6.d without some review, so it's fine for now.
bartolin ok, I'm fine with that
Frameless The rest of the tests are all to check for bug regression. 17:21
(in that file)
[Coke] added github.com/rakudo/rakudo/wiki/cont...inancially 17:44
lucs \w5 17:45
[Coke] added some to github.com/rakudo/rakudo/wiki/rt-contributing 17:52
bugfolks: feel free to start fleshing those out. 17:53
I will continue to drop more stuff in there that I've been keeping in gists in the past. 17:54
perlpilot Frameless++ just read your post. 17:58
dalek ast: dbdc612 | usev6++ | S (3 files):
Unfudge two passing test on JVM

also use 'todo' instead of 'skip' for on precomp test on Moar
22:18