Geth roast/6.c-errata: 0940d21cf4 | (Zoffix Znet)++ | packages/Test/Util.pm
Update Test::Util version in 6.c-errata

Need it for 6.c-only tests I'm gonna move from master		 00:16
Zoffix 6.c-errata failures: t/spec/S04-phasers/end.t t/spec/S17-supply/basic.t t/spec/S29-context/exit.t t/spec/S32-str/sprintf.t 00:23
(I hope none are from my change to de-specialize `open "-"`
)
Geth roast/6.c-errata: cba73db7d0 | (Zoffix Znet)++ | S32-io/open.t
Move open "-" specialness tests to 6.c

They were only added to master branch and we're keeping this behaviour only in 6.c lang, so it makes sense to keep them only in the 6.c-errata branch.		 00:26
rakudo/nom: 03b1febc93 | (Zoffix Znet)++ | t/spectest.data.6.c
Add S32-io/open.t to 6.c test files
 00:27
Zoffix wtf 00:44
This passes the test: gist.github.com/zoffixznet/3320ed4...c961f4d43d 00:45
But if put into t/spec/S32-io/open.t, it gets stuck waiting for input on STDIN :/ 00:46
even tho it got "use v6.d.PREVIEW;" at the top :
and if I add "diag $*PERL.version ;" before the tests in the file it prints 6.d :S 00:50
ZofBot: what's happening?
ZofBot Zoffix, Note however that any dispatcher derived from "proto" will map named arguments to known declared positional parameters and call the "multi" candidates with positionals for those arguments rather than named arguments
Zoffix huh 00:52
MasterDuke Zoffix: no idea, but shouldn't the description of the second test be '&open(IO::Path)'?
Zoffix And afetr the .t file hangs, if I run the file with code with git it hangs too, but if I kill it and re-run then it works. 00:53
MasterDuke: yeah, thanks
*sigh* 00:57
Gonna debug further tomorrow
Geth rakudo/unspeshul-open: 76359abea7 | (Zoffix Znet)++ | src/core/IO/Handle.pm
Make open("-") non-special in 6.d
 00:58
roast/unspeshul-open: 481ec90646 | (Zoffix Znet)++ | S32-io/open.t
Test open("-") is not special in 6.d

The specialness tests were moved[^1] to 6.c-errata. Switch entire file to 6.d and test that "-" is not special
  [1] github.com/rakudo/rakudo/commit/03b1febc93		 00:59
Zoffix s:g/hangs/waits for STDIN/; 01:00
Something isn't propagating the language right or maybe nqp::getcomp('perl6').language_version isn't the right way to get it at all times and need to use $*PERL.version 01:01
travis-ci Rakudo build failed. Elizabeth Mattijsen 'We want the definite type, jnthn++' 01:23
travis-ci.org/rakudo/rakudo/builds/280938056 github.com/rakudo/rakudo/compare/7...2ab751b838
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
TimToady AlexDaniel: by that argument it should be called !lang 01:24
m: <perl5 golang c# !lang 6lang ruby>.sort.say
camelia (!lang 6lang c# golang perl5 ruby)
AlexDaniel unless "…4langs already exist, no 01:25
…5 I mean. O-ops :) 01:26
it's not that the proposal was motivated by this
simply it's one of its benefits 01:27
Zoffix Clearly it should be "\0lang" :P 01:28
Zoffix & depowers 01:29
AlexDaniel Zoffix: no, because this new language named \0 will be first then. 01:30
Zoffix Dammit, can't win :) 01:31
travis-ci Rakudo build passed. Zoffix Znet 'Polish unicode -> ascii op converter 02:08
travis-ci.org/rakudo/rakudo/builds/281002994 github.com/rakudo/rakudo/compare/a...f2f98f796b
TimToady so if the real name of "golang" is Go, the real name of 6lang is "6"? 02:25
MasterDuke "p6lang. with a 'p', like in psychology" 02:32
ugexe i like bloop blorp 02:35
samcv bloob blorp? 02:49
travis-ci Rakudo build canceled. Stefan Seifert 'Remove unnecessary creation of an extra serialization context 03:00
travis-ci.org/rakudo/rakudo/builds/280445160 github.com/rakudo/rakudo/compare/9...9fe7dbbd7b
buggable [travis build above] ✓ All failures are due to: missing build log (1 failure).
Geth roast/6.c-errata: 947ba7960e | (Samantha McVey)++ | S02-literals/quoting-unicode.t
Remove ornate parenthesis from quoting-unicode.t

These should never have been added as they are not considered brackets by Unicode. There are no modules in the ecosystem which use it and saving this for Perl 6.d would not be viable or needed.		 03:53
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Move up Rakudo::Internals::CompilerServices 05:10
travis-ci.org/rakudo/rakudo/builds/281069472 github.com/rakudo/rakudo/compare/6...63f898f67e
nine .tell Zoffix Please do not use constructs like "$*PERL.version after v6.c". That it does what you need it to do is a plain bug that really needs to be fixed. You are trying to get your caller's language version by accessing something that should be different only in lexical scope. $*PERL.version behaves completely bogus and unpredictable. 05:48
yoleaux nine: I'll pass your message to Zoffix.
nine .tell Zoffix To make it clear: there is currently _no_ way at all to correctly do what you are trying. That's to me the single biggest blocker of 6.d. 05:50
yoleaux nine: I'll pass your message to Zoffix.
travis-ci Rakudo build passed. Zoffix Znet '"Remove" is_approx in 6.d 06:42
travis-ci.org/rakudo/rakudo/builds/281090613 github.com/rakudo/rakudo/compare/7...043f2ae4cb
[Tux] Did the yesterday timings do a wake-up call? Drumrollllllll 06:56
This is Rakudo version 2017.09-154-g03b1febc9 built on MoarVM version 2017.09.1-50-g3059ba28 06:57
csv-ip5xs 1.351 - 1.461
test 10.236 - 10.898
test-t 3.247 - 3.318
csv-parser 13.066 - 13.113
TimToady finally back home (we had a 10-hour layover in YVR) and now will collapse into bed 07:03
nine csv-ip5xs still looks slower than it used to be? 07:13
TimToady and, of course, the pipe under the guest bathroom sink immediately breaks and floods everything before ever we get into bed... 07:19
gfldex TimToady: that broken pipe will still be there in the morning 07:20
TimToady hopefully the water won't...
gonna run a fan overnight to help guarantee that... 07:21
travis-ci Rakudo build passed. Zoffix Znet 'Deprecate Str.lines: :$count in 6.d' 07:23
travis-ci.org/rakudo/rakudo/builds/281093901 github.com/rakudo/rakudo/compare/c...d4939c38db
lizmat [Tux] that's less than I was going for :-( 07:25
yoleaux 28 Sep 2017 22:39Z <dogbert2> lizmat: eb9c3d4dd7791ad1 fails
stmuk has anyone tried building on the new OS X (High Sierra) with their new fs? 08:07
travis-ci Rakudo build passed. Zoffix Znet 'Remove $*MAIN-ALLOW-NAMED-ANYWHERE 08:10
travis-ci.org/rakudo/rakudo/builds/281101058 github.com/rakudo/rakudo/compare/0...b4b167f5e0
Geth rakudo/nom: 4959df3f31 | (Elizabeth Mattijsen)++ | 2 files
Remove $/ support from CompilerServices

This reverts the first commit (and subsequent commits related to this) that dogbert++ found to be the first that caused failures for him.		 08:19
lizmat stmuk: not me 08:20
travis-ci Rakudo build passed. Zoffix Znet 'Add S32-io/open.t to 6.c test files' 08:57
travis-ci.org/rakudo/rakudo/builds/281109987 github.com/rakudo/rakudo/compare/9...b1febc93ab
Zoffix . 08:58
yoleaux 05:48Z <nine> Zoffix: Please do not use constructs like "$*PERL.version after v6.c". That it does what you need it to do is a plain bug that really needs to be fixed. You are trying to get your caller's language version by accessing something that should be different only in lexical scope. $*PERL.version behaves completely bogus and unpredictable.
05:50Z <nine> Zoffix: To make it clear: there is currently _no_ way at all to correctly do what you are trying. That's to me the single biggest blocker of 6.d.
Zoffix interesting...
Well, at least that explains the bug I was having :D 08:59
Zoffix goes to sleep again
DrForr takes notes as well. 09:01
Zoffix TimToady: the real name of 6lang is 6lang :) 09:06
Or "slang"
Or its alternative name "Perl 6"
Unless you plan to entirely replace the name, which I'm all for, but I know some other people aren't :) 09:07
DrForr As long as the 6 faces in the right direction :) 09:09
Zoffix Yup :) 09:10
DrForr Being able to encaase blocks in 'for False, True -> $*PURE-PERL { .. }' makes for easy rewriting of test suites... 09:12
The more I do this sort of thing the more an outline for Perl 6 best practices comes to mind..
Geth rakudo/nom: 22d3d933b3 | (Zoffix Znet)++ | 2 files
Revert "Deprecate Str.lines: :$count in 6.d"

This reverts commit 01d4939c38dba9c98b76991cd59be9e438d42a76.
  irclog.perlgeek.de/perl6-dev/2017-...i_15233995		 09:16
rakudo/nom: 8ed7adf1a6 | (Zoffix Znet)++ | 2 files
Revert ""Remove" is_approx in 6.d"

This reverts commit cd043f2ae4cb617eae89252de3263a49c54311f3.
  irclog.perlgeek.de/perl6-dev/2017-...i_15233995
rakudo/nom: 68fdeff3b6 | (Zoffix Znet)++ | t/02-rakudo/v6.d-tests/01-deprecations.t
Revert "[6.d] Deprecate IO::Handle.slurp-rest"

This reverts commit 3341384bfe1341200a75ddc7ec869812cd58aeed.
  irclog.perlgeek.de/perl6-dev/2017-...i_15233995		 09:18
rakudo/nom: 142f772e32 | (Zoffix Znet)++ | src/core/IO/Handle.pm
Revert "[6.d] Deprecate IO::Handle.slurp-rest"

This reverts commit 3341384bfe1341200a75ddc7ec869812cd58aeed.
  irclog.perlgeek.de/perl6-dev/2017-...i_15233995		 09:19
Zoffix damn now conflicts
Geth rakudo/nom: a65d5f922a | (Zoffix Znet)++ | 3 files
Revert "[6.d] Deprecate IO::Path.chdir"

This reverts commit 6d2adb20f2529e36e32b50936281ef3f11f078b0.
  irclog.perlgeek.de/perl6-dev/2017-...i_15233995
Conflicts: t/02-rakudo/v6.d-tests/01-deprecations.t		 09:21
rakudo/nom: 44d5256cd2 | (Zoffix Znet)++ | 3 files
Revert "Deprecate dummy arg on .Rat/.FatRat…"

This reverts commit 4c337e8ef9fa8a117761f5a74dc444a188471b71.
Conflicts: t/02-rakudo/v6.d-tests/01-deprecations.t		 09:22
Zoffix ZOFVM: Files=1275, Tests=152181, 142 wallclock secs (20.72 usr 3.14 sys + 3005.78 cusr 208.19 csys = 3237.83 CPU) 09:29
Geth 6.d-prep: 9913c2aab4 | (Zoffix Znet)++ | TODO/FEATURES.md
List TODO: need a way to get lang version
nine
.oO(Zoffix doesn't seem to sleep very long. Or long at all. Or more than a moment or two.)
 09:43
Zoffix :)
I changed my mind on sleeping :)
Geth rakudo/nom: d3c481854e | (Elizabeth Mattijsen)++ | src/Perl6/World.nqp
Improve readability of BUILDALL autogenerating logic

And perhaps a bit of streamlining, but I doubt that'd be noticeable.		 09:46
[TuxCM] lizmat, going down ... 11:09
This is Rakudo version 2017.09-162-gd3c481854 built on MoarVM version 2017.09.1-50-g3059ba28
csv-ip5xs 1.260 - 1.265
test 9.359 - 9.486
test-t 3.183 - 3.208
csv-parser 12.003 - 12.101
lizmat ah, better :-) 11:10
was hoping to get below 3 though :-(
m: say 3.183 / 1.260
camelia 2.526190
lizmat only 2.5x slower than XS + Inline::Perl5 11:11
nine
.oO(for low iteration counts)
 11:12
timotimo you think we should start doubling the amount of loop iterations? :)
Geth rakudo/nom: 371befe8f7 | (Elizabeth Mattijsen)++ | src/Perl6/World.nqp
Give @/% attributes optimized handling

  - instead of having to go through p6store, which then does a .STORE
  - we call .STORE directly
  - This makes them a few percent faster		 11:13
nine At Frame 38, Instruction 88, op 'prepargs', operand 0, MAST::Local of wrong type (8) specified; expected 7 11:16
trying to build nqp with current Moar master
timotimo need a full reconfigure & rebuild? 11:17
nine tried make distclean and Configure.pl but no change 11:18
timotimo huh, interesting
nine neither does realclean
Ah, stupid me. I just forgot the git pull 11:19
timotimo heh :)
so a bump is missing?
nine No, it's the other way round. Old nqp doesn't like new MoarVM 11:21
timotimo oh
travis-ci Rakudo build passed. Zoffix Znet 'Revert "Deprecate dummy arg on .Rat/.FatRat…" 11:23
travis-ci.org/rakudo/rakudo/builds/281216589 github.com/rakudo/rakudo/compare/a...d5256cd2cb
lizmat nine: re test-t and more iterations: doubling the iterations doubles the runtime for me (minus startup overhead) 11:24
nine lizmat: csv-ip5xs spends ~ half a second on startup. That's a huge part of the 1.2s 11:25
Here it's pretty close to 1s for 10000 iterations and 5s for 100000 iterations
timotimo oh, the benchmark tux runs subtracts startup, but perhaps the nativecall stuff isn't all set up during an empty workload 11:30
lizmat $ time perl6 -MInline::Perl5 -e '' # 0.243 for me, nine 11:32
could it be that the additional .25 second is needed to parse the Perl 5 modules ? 11:33
nine If it subtracts startup, the machine must be quite slow :) Considering that time perl6 -Ilib test-t.pl < /tmp/hello.csv runs in real 0m2.844s here
timotimo generated nativecall subs are in nom already, aren't they?
lizmat timotimo: not sure what you mean 11:34
nine lizmat: that's missing a lot of overhead. You get closer by letting it run a single iteration and measuring that.
Takes 0m0.537s here
timotimo nativecall generates subs much like the buildallplan now does, but only when the sub actually gets invoked
i.e. you only pay for the subs you call, but if you measure startup based on "nothing", you won't call most of the subs 11:35
how far off the mark am i?
nine That ^^^ and loading the native library itself, which is only done on the first call.
timotimo that's probably a bunch cheaper 11:36
nine And starting Perl 5 (which is fast, granted) and loading the Perl 5 modules
And loading the modules triggers a lot of work in Inline::Perl5 for setting up the wrapper classes.
time perl6 -MInline::Perl5 -e '' takes 0.190s here, time perl6 -MInline::Perl5 -e 'use Text::CSV_XS:from<Perl5>;' takes 0.461s 11:38
Zoffix buggable: speed 4 11:42
buggable Zoffix, ▄▂▁▂ dates: 2017-09-29–2017-09-29; range: 3.183s–3.318s; speed: 3% faster
Zoffix nine: BTW is the $*PERL.version thing an unsolved problem or is there a known path to follow that just needs someone to do it? 11:43
Was just wondering if I had any chance to fix whatever the issue is
Or `nqp::getcomp('perl6').language_version()`
dogbert17 lizmat: so you only had to revert parts of your work not all of it ?
lizmat if it now works for you, yes
timotimo we've discussed - or tried discussing? - how to get the caller's perl version
at the swiss perl workshop 11:44
Zoffix timotimo: was any of it written down?
lizmat timotimo: we need something like $/ that indicates the Perl version of the current scope
timotimo "like $/"? 11:45
you mean a variable accessible through CALLER:: or CLIENT:: ?
lizmat yeah, like we handle methods setting / inspecting $/ in their outer scope
dogbert17 is at $work so has to download the doc repo 11:46
lizmat because just looking at the callers' package version is not enough
dogbert17 lizmat: the error from yesterday seems to be gone, doing a complete run just to see that nothing else crops up 11:53
Geth 6.d-prep: 9f0cccb3fd | (Zoffix Znet)++ (committed using GitHub Web editor) | TODO/FEATURES.md
Fix typo; MasterDuke++
 12:01
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Improve readability of BUILDALL autogenerating logic 12:05
travis-ci.org/rakudo/rakudo/builds/281224272 github.com/rakudo/rakudo/compare/4...c481854ec3
dogbert17 lizmat: htmlify.p6 is happy again 12:12
lizmat cool :-)
and hopefully a bit faster :-)
dogbert17 and I see that [Tux] benchmark runs faster, lizmat++ 12:13
we're closing in on 3 seconds
Zoffix buggable: speed 100 :4 12:18
buggable Zoffix, ↑↑↑↑↑↑↑↑↑ ▂ ▅ █ ▂ dates: 2017-08-18–2017-09-29
Zoffix, █████████▁█▄▃ ▅ ▄ ▄ ▅ █ █▆ █▃ range: 3.183s–4.592s
Zoffix, █████████████▆█▂▃▆▃▃▅▆▃▂ █▇▁▂ ▂▂ ▂ ▁█▄▂▂▄█▇█▅▂▃▃▂▆▁▃ ▄▇▅▄██▃▃▂▃ ▁▂ ▂ ▂ ▄ ▁ ▃▄ ▄ ▄██▅ speed: 38% faster
Zoffix, █████████████████████████▇█████████▆▇█████████████████▇▇██████████▇▇██▇█▇▇██▇█▇█▆▇▇▇▇██████▄████▅▃▁▂
Zoffix passes around a round of ++
buggable: speed 120 :4 12:28
buggable Zoffix, ↑ █▇▆↑↑↑██▇██↑↑↑↑↑ ▁ ▄ dates: 2017-08-04–2017-09-29
Zoffix, █ ▇ ████████████████ ▅▁ ▃ ▂ ▂ ▂ █ █▃ ▆ range: 3.183s–4.925s
Zoffix, █▆▅▆▅▇█▇▅▆▆▃▃████████████████▆███▄█ ▁▄▂▂▃▄▂ █▅ ▁ █▂ ▂█▅█▃ ▂▁ ▄ ▂ ▂▅▃▃██▂▂ ▁ ▃ ▁▂ ▂ ▃██▃ speed: 43% faster
Zoffix, ████████████████████████████████████████████▇▆██▇█▇█▇▇█▅▆▇██████████████▇█▆▆██████████▆▆▇█▆█▆▆█▇▆█▆▇▅▆▆▇▆▇██▇█▇▄████▄▂▁▂
Zoffix buggable: speed 115
buggable Zoffix, ▄▆▄▄▄▄▃▃██▇↑↑↑█████↑↑↑↑↑▄▆▅▄▃▅▃▃▄▃▃▃▄▃▃▂▂▅▄▃▃▂▃▃▂▃▂▂▃▅▃▃▃▃▅▄▆▃▃▃▃▃▃▃▃▂▂▃▄▃▃▇▅▃▃▃▃▂▂▃▃▂▃▂▂▃▂▂▃▂▃▂▂▂▂▂▂▃▃▂▃▂▂▃▆▄▃▂▁▁▁ dates: 2017-08-06–2017-09-29; range: 3.183s–4.925s; speed: 17% faster
Zoffix
.oO( the "speed: 17% faster" thing needs to do some sort of averaging :/ )
 12:29
Geth rakudo/nom: 9d4a833b83 | (Elizabeth Mattijsen)++ | src/Perl6/World.nqp
Take all constant SVal values out of the loop

Should help a bit with CPU and memory pressure of compilation.		 12:36
[Coke] (stuffed bear, rubber duck) also the coworker who really doesn't care about your problem but is a good sport and listens to you explain it so y---ah, that's what i did. 12:48
(I also have a literal rubber duck on my desk)
DrForr I used to have a Klingon teddy bear for such a purpose.
moritz with two hearts? 12:50
Geth rakudo/nom: 6902c59042 | (Elizabeth Mattijsen)++ | src/Perl6/World.nqp
Turns out callmethod also takes a :name

So we don't need these SVals after all.		 12:51
DrForr Never listened.
moritz DrForr: sounds like a tragic story 12:52
timotimo oh jeez 12:56
DrForr It's probably in storage, actually. 12:57
timotimo imgur.com/5YhyQmE - control flow graph bug from hell
moritz alone, forgotton and abandoned
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Give @/% attributes optimized handling 12:58
travis-ci.org/rakudo/rakudo/builds/281250158 github.com/rakudo/rakudo/compare/d...1befe8f7aa
DrForr apt update - what hath you wrought, you evil bastard... 13:04
lizmat afk until later& 13:26
stmuk ohshitgit.com/ 13:53
timotimo that's a cool format 13:56
Zoffix "I tried to run a diff but nothing happened?!" oh cool... I always hated that. Nice to know there is a way to see it 13:57
jdv79 to whom is git hard?
travis-ci Rakudo build errored. Elizabeth Mattijsen 'Take all constant SVal values out of the loop 13:58
travis-ci.org/rakudo/rakudo/builds/281276381 github.com/rakudo/rakudo/compare/3...4a833b8322
buggable [travis build above] ✓ All failures are due to: timeout (1 failure).
jdv79 how come test runs are so fragile? 14:02
Zoffix it's a free service? 14:03
jdv79 so is cpantesters 14:05
thats a lame excuse
in short, we can do better
:)
Zoffix "we"? We don't own travis
jdv79 who needs travis? 14:06
Zoffix why is the sky blue?
ZofBot: what's the question game?
ZofBot Zoffix, csv-ip5xs 1.339 - 1.341
jdv79 not equivalent but ok
ilmari jdv79: people who want to test things that aren't cpan modules 14:07
jdv79 is travis the best we can do? 14:09
timotimo for free, yeah
jdv79 no. challenge accepted. 14:10
timotimo wow, ok
Zoffix heh 14:13
jdv79 do we have a doc pointing to all testing infra? 14:14
i feel like i asked this not that long ago.
Zoffix m: for (<a b c>, <d e>).map({Slip.new: $_ }) { dd $_ } 15:02
camelia $("a", "b", "c")
$("d", "e")
Zoffix m: for (<a b c>, <d e>).map({.Slip}) { dd $_ }
camelia "a"
"b"
"c"
"d"
"e"
Zoffix doesn't get Slips
m: for (<a b c>, <d e>).map({Slip.new: |$_ }) { dd $_ } 15:04
camelia "a"
"b"
"c"
"d"
"e"
Zoffix ah right doh >_<
jnthn omg 15:07
Zoffix ?
jnthn Been hunting bugs in IO::Socket::Async::SSL
First, it turns out a fix I attmepted and thought didn't work earlier actually did work, but a second bug inside the failing test made it look like that fix didn't work 15:08
Anyway, fine
But that left me with an occasional hang
Which looks like it may be because DESTROY can fire at pretty much any point 15:09
Including the least possible convenient one
And cause a circular lock contention
Zoffix :o
jnthn That took some finding. 15:11
One thread held the lock used to protect libssl against concurrent use and wanted to keep a Promise. The other was adding a then to the Promise, so holding the internal lock on the Promise, but GC ran during Promise construction, ran a DESTROY, which tried to do a lock acquisition of the lock used the protect libssl. 15:12
stmuk is 6.c-errata supposed to still pass on previous rakudo releases (eg. last month's)? 15:13
Zoffix m: with (1, 2, 3).Seq -> $s { $s.skip for ^1000 } 15:14
camelia Method 'pull-one' must be implemented by Iterator because it is required by roles: Iterator.
in block at <tmp> line 1
in block <unit> at <tmp> line 1
jnthn hurrah, t/thread-stress.t ran and passed 100 times without a hang
Or crash
Or wrong answer
timotimo greetings jnthn
jnthn o/ timotimo
Zoffix stmuk: not necessarily, since wrong tests testing incorrect behaviour can get fixed and older rakudo versions might still have the incorrect behaviour 15:15
the tarball gets shipped with the 6.c-errata snapshot and that version is supposed to pass for that rakudo version 15:18
stmuk Zoffix: OK that's what I suspected.
I can't see anyway of telling easily what git version of 6.c-errata probably passed for 2017.09 other than guessing from the date and comparing what's in the tarball 15:25
Zoffix What the reason for you wanting to tell that? 15:26
stmuk I'm wondering whether there should be a VERSION file or similar created in the roast directory 15:27
or even tagging
Zoffix What for? 15:28
stmuk so that someone can use git to debug problems rather than having to use the tarball 15:29
Zoffix doesn't follow 15:30
I'd think all the problems would be debugged pre-release.
stmuk what about trying the last rakudo release on a OS release which didn't exist at the time of the last release? 15:31
Zoffix Get the tarball :) You'll get both the code and the tests. If you want to make a patch fixing something, then build from master ¯\_(ツ)_/¯ 15:32
stmuk it just seems strange and restricting there is information stored outside of git .. particularly given the importance of roast as defining what perl 6 is 15:33
Zoffix It'd probably not be a VERSION file in roast, but a rakudo/rakudo/tools/build/ROAST_VERSION file + tag in roast 15:34
stmuk yeah something along those lines maybe
Zoffix stmuk: it makes sense tho. The older rakudo version does not implement the language correctly, so the fixed test fails now. 15:35
The whole tagging thing just adds micro language releases "this is rakudo that passes language version 6.c-errata-g.312321.4123fa3" 15:36
oh crap. 15:37
stmuk isn't that what you have with monthly releases anyway?
Zoffix m: my $s := (1, 2, 3); say $s.skip; say $s 15:38
camelia (2 3)
(1 2 3)
Zoffix m: my $s := (1, 2, 3).Seq; say $s.skip; say $s
camelia (2 3)
(2 3)
Zoffix different semantics :(
stmuk: yeah, ok, I guess micro-releases for 6.c-errata do make sense. 15:39
And the real reason it's crap is 'cause of this: 15:40
m: my $s := (1, 2, 3).Seq; $s.cache; say $s.skip; say $s 15:41
camelia Method 'pull-one' must be implemented by Iterator because it is required by roles: Iterator.
in block <unit> at <tmp> line 1
Zoffix It's broken there, but in my fix, attempting to .skip a cached Seq doesn't toss the element from the Seq anymore
cpan@perlbuild4~/CPANPRC/rakudo (nom)$ ./perl6 -e 'm: my $s := (1, 2, 3).Seq; $s.cache; say $s.skip; say $s'
(1 2 3)
cpan@perlbuild4~/CPANPRC/rakudo (nom)$ ./perl6 -e 'm: my $s := (1, 2, 3); say $s.skip; say $s' 15:42
(2 3)
(1 2 3)
nine Regarding Travis' instability: I do have a couple gigs of RAM and lots of storage free on my server (even CPU cycles) and could easily set up a VM for running Travis, Jenkins or whatever. We'd just need someone tending to that installation. 15:44
We don't _have_ to download and compile even Perl 5 for every single test run after all...
Zoffix nine: OK, let's set it up. Would be helpful to run stresstest as part of the test run as well 15:50
nine Sooo....what should I call that VM? 15:51
rakudo-ci it is
Zoffix m: say ('a'..'z').pick xx 7
camelia (j n v s k h u)
Zoffix m: say ('a'..'z').pick xx 7
camelia (j f e g n j a)
Zoffix :)
stmuk metacpan.org/pod/App::test::travis looks interesting 15:55
Zoffix Filed .skip issue as rt.perl.org/Ticket/Display.html?id=132184 16:08
SlippyIterator is a user class? 16:38
*role
ilmari read that as SloppyIterator...
ugexe i like my iterators extra sloppy 16:44
AlexDaniel Zoffix++ # being OK about changing his mind. I wish there was more of it around here. Me included. 16:50
.oO( life is hard :S )
 16:51
. 16:52
japhb: thanks :) .tell didn't work but I did see it 16:53
m: say 42 16:54
evalable6 42
AlexDaniel loves this little feature
Zoffix: re ｢git diff --staged｣, it's actually part of my normal workflow 16:58
so I always look at what I'm about to commit before running ｢git commit｣
… and then I also ｢git diff HEAD^..HEAD｣ after committing, better safe than sorry… :)
bonus tip: ｢git diff --staged｣ does not take much effort to type because you can Ctrl-R “sta” Enter 16:59
assuming that you've used it in the past 17:00
another bonus tip: remove bash_history limits so that you have infinite command history. Now you only have to type any command *once* in your lifetime. Bring your .bash_history file to any new system you use.
Zoffix AlexDaniel: changing mind about what? 17:02
`git diff HEAD^..HEAD` I think `git show` does something like that 17:03
AlexDaniel Zoffix: well, I was reminded about it when I saw “I changed my mind on sleeping”, but I've seen you go “ok, I'm switching my position” several times now 17:04
Zoffix Ah
I have aliases: `gd` for `git diff` and now `gds` for `git diff --staged` :)
`grr` for 'git pull --rebase' and `GRRR` for `git reset --hard; git pull --rebase` 17:05
Zoffix grrrs alot :P
AlexDaniel that's another way to do it, yeah. But Ctrl+r is universal in a way that you can quickly get *any* command as long as you remember something unique about it 17:06
japhb alias glr='git log -p --reverse'
I like that one, but even more I use 'gpr', which is ... hold on a sec while I pull up my aliases ...
MasterDuke alias gg='git grep -E' 17:07
use that a lot
japhb gpr () { cur=`git rev-parse HEAD`; git pull; git log -p --reverse $cur...; }
AlexDaniel MasterDuke: have you tried ack? :)
it's not as fast though
japhb That gpr is command is basically what I use to keep up with every Perl 6 repo I follow
MasterDuke i think i have it installed, but never use it 17:08
jnthn When I was teaching git I used to have git god as git log --graph --oneline --decorate :)
yoleaux 16:18Z <MasterDuke> jnthn: not to distract from concurrency stuff, but when you've got some time, any thoughts on github.com/MoarVM/MoarVM/pull/689?
japhb That and keeping '/^commit ' in my less history :-)
jnthn: glg () { git log --graph --pretty=format:'%Cred%h%Creset -%C(yellow)%d%Creset %s %Cgreen(%cr) %Cblue<%an>%Creset' --abbrev-commit --date=relative "$@" | less -R ; }
jnthn japhb: Neat :) 17:09
japhb Huh, wonder why the color is missing when running that on my work computer 17:10
timotimo no need to |less that 17:12
git gives you a pager for free
mst unless you want to control which pager
timotimo hm, GIT_PAGER maybe? :) 17:13
japhb adds --color=always to the glg command, since that's rather the point 17:18
timotimo: Because I try to reduce the number of different ways I am configuring things, so that I can e.g. copy my .bash_aliases to a new computer, source it, and be immediately more productive, even before I copy more stuff over or if I forget to do so. 17:20
It costs me nothing to be explicit about what I want in my .bash_aliases, plus I can copy and paste from there to show people things, and not have to explain "Oh, and make sure you've got color enabled in your git config and all"
timotimo fair enough 17:24
Zoffix nine: so should I set everything up for rakudo-ci? Or are you doing something already 17:39
nine Zoffix: I've only installed the very basics so far. What _do_ we actually want to run? 17:40
Zoffix No idea :)
I mean, I have an idea of the end goal: something that builds new commits, stresstests them, and reports when failures happen. 17:41
But I guess we'd still have travis running, since it also does OSX? At least the time it takes for it to complete can be halved
Oh and it doesn't stresstests anything, just `make test` 17:42
nine Oh, Travis is not actually really free software? 17:43
Zoffix I don't know.
nine Wikipedia says: Although the source is technically free software and available piecemeal on GitHub under permissive licenses, the company notes that it is unlikely that casual users could successfully integrate it on their own platforms.[
We do have our own Jenkins at work. I'm not a massive fan however. And apparently it's very difficult/impossible to have it test random branches 17:44
Zoffix Well, I already have an IRC bot that can cut releases. I can just modify it slightly to just be a testing bot.
nine I wonder how many negative points that will give us in terms of being a sustainable free software project ;) 17:46
Zoffix ZofBot: what are these points? 17:48
ZofBot Zoffix, For instance: sleep& 6model The "new object model"/"MOP" that's currently in rakudo
Zoffix
.oO( 6model... we already have a 6lang-ish term in use... )
nine But I can't find that sheet anymore that listed the factors
Zoffix I wonder if the sheet was invented by people pushing corporate version of travis/github/etc :P 17:49
Need to nuke and rebuild R6 db from scratch.... 18:07
ZofBot: in 3... 2... 1...
ZofBot waves good night.
Zoffix \o/ 18:08
R6 = fail.rakudo.party
Nuking completed 18:37
buggable: tag 18:54
buggable Zoffix, Total: 1648; 6.D: 1; 9999: 10; @LARRY: 24; ANNOYING: 7; BOOTSTRAP: 5; BUG: 603; BUILD: 12; CONC: 48; DOCS: 1; EXOTICTEST: 1; GLR: 3; IO: 23; JVM: 53; LTA: 177; MATH: 4; META: 1; MOAR: 1; NATIVECALL: 22; NYI: 55; OO: 13; OPTIMIZER: 8; OSX: 2; PARSER: 6; PERF: 28; POD: 19; PRECOMP: 14; REGEX: 50; REGRESSION: 37; REPL: 6; RFC: 63; RT: 1;
Zoffix, SECURITY: 2; SEGV: 25; SINK: 1; SITE: 1; SPESH: 1; STAR: 7; TESTCOMMITTED: 10; TESTNEEDED: 45; TODO: 13; UNI: 26; UNTAGGED: 414; WEIRD: 2; WINDOWS: 3; See fail.rakudo.party/ for details
Zoffix Fixed it getting cut off
buggable: tag security
buggable Zoffix, There are 2 tickets tagged with SECURITY; See fail.rakudo.party/t/SECURITY for details
Zoffix
.oO( remove .EVAL method )
 18:55
Fixed both RT#128684 and that eval-in-regex thing I posted somewhere 18:56
nine: geth is ded 18:59
I merged github.com/rakudo/rakudo/pull/1175 "Move security RT#131079 fix from Grammar to Actions " 19:00
man synopse bot is also ded...
ZofBot: is it a botstrike?
ZofBot Zoffix, dothat): 1,2,3 or $obj
Zoffix no idea how synopse bot is meant to be started... don't see a service for it and neither 6lang nor perl scripts in /home/synopsebot start up due to missing modules 19:04
MasterDuke timotimo: ^^^
timotimo there's a system file 19:05
● synopsebot.service - An IRC bot called "synopsebot" for links to the design documents. 19:06
Zoffix oh dam
It wasn't showing up on `sudo service sy[TAB]` 19:07
mst that can happen because the tab complete is running with your privileges 19:13
Zoffix Ah 19:14
timotimo anyway, that bot wants rewriting with the nicer irc bot library, and also to learn a thing or two, like not spitting out the same link multiple times over the course of two minutes
Zoffix OK, I'll rewrite it tonight
timotimo m: my $x = "abc"; $x [R~]= "xyz"; say $x 19:22
camelia Potential difficulties:
Useless use of [R~]= in sink context
at <tmp>:1
------> 3my $x = "abc"; $x 7⏏5[R~]= "xyz"; say $x
xyzabc
timotimo ^- sinkbug
Zoffix fileit :D 19:24
timotimo there might already be one
also i'd like to golf it a bit first
looks like it happens for any metaop with brackets before an assignment metaop 19:25
Zoffix looks at extra correspondence after RT#132179 was closed 19:26
s/Texas/ASCII/ sounds good to me. So.. should be do it?
synopsebot6: weren't you supposed to link to tickets?
#132179 RT#132179 132179 19:27
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=132179
Zoffix RT#132179
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=132179
Zoffix :| now you're just screwing with me
Oh, it doesn't understand ACTION
timotimo oh 19:30
rt.perl.org/Public/Bug/Display.html?id=131251 19:32
Zoffix Opened a doc Issue for the s/Texas/ASCII/ RFC: github.com/perl6/doc/issues/1584
timotimo this is the bug for metaop assign useless use bug
it should also have info for X and Z in there
travis-ci Rakudo build errored. Elizabeth Mattijsen 'Takes some more constant QAST out of the loop' 19:33
travis-ci.org/rakudo/rakudo/builds/281401195 github.com/rakudo/rakudo/compare/6...4eb666b0b5
buggable [travis build above] ☠ All failures are due to: timeout (1 failure), failed make test (1 failure). Across all jobs, only t/04-nativecall/06-struct.t test file failed.
Zoffix TimToady: AlexDaniel some of Damian Conway's thoughts on the language naming: gist.github.com/zoffixznet/522abeb...afeebc058a 19:51
.ask nine would I have the option to getting keys to geth so I could restart it, or pull new changes to it, if needed? 22:48
yoleaux Zoffix: I'll pass your message to nine.
Zoffix it's ded right now and IIRC it don't got latest fixes from the repo.
ZofBot: looks like it's just you and me here. Let's make a baby 23:03
ZofBot Zoffix, Link: rt.perl.org/rt3/Public/Bug/Display...?id=131079
Zoffix Yup. You're thinking what I'm thinking. Let's make a ticket bot
stmuk liuqu / liudao 23:30
Zoffix "Six districts / stay" ? 23:31
stmuk I had six states 23:36
rokudo is "liudu" (these are mostly Buddist terms with esoteric and multiple meanings) 23:39
Zoffix Hacktoberfest Issues: hacktoberfest.digitalocean.com/#projects 23:43
stmuk also Rokushu 23:45
6 something or other .. I guess there is danger in naming similar to waking up with a mispelt tattoo spelling something different to what you wanted 23:47
Zoffix :)
BenGoldberg I've got a proposal to make! I think we should rename perl6. It should be called "pie". I like pie! Especially pumpkin. Who likes pie? :) 23:51
PS: ^ not a serious suggestion, /me is suffering a case of the sillies.
Zoffix I like pie!
Zoffix registers pie.party
"getlex: outer index out of range" nice... :/ 23:53
Something with precomping constant regexes 23:58
m: with '/tmp/ZZZZRT'.IO.mkdir { .add('Foo.pm6').IO.spurt: q|unit class Foo; constant $RT_RE = rx/« [RT]?'#'? <( <[0..9]>**{5..6} »/; method x ($e where $RT_RE) {say "in"}| }
camelia ( no output )
Zoffix m: use lib </tmp/ZZZZRT>; use Foo; Foo.new.x: "RT#131417"
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131417
camelia getlex: outer index out of range
in regex at /tmp/ZZZZRT/.precomp/B605C1056356D83F4CB89430E4824F7C7F6DEFDE.1506729021.93654/20/201A6B3053CC1422D2C3670B62616221D2290929 line 1
in method x at /tmp/ZZZZRT/Foo.pm6 (Foo) line 1
in block <unit> at…

Powered by Cro and the Raku® Programming Language.

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

Thank you!