dalek kudo/nom: b7f08d7 | (Zoffix Znet)++ | docs/ChangeLog:
Populate ChangeLog

Add changes from commits 9180464, 23daf3b, ac19874, 027dd39, 0295443, 6d0cc07, 29a1107, 94b52c1, 959cd3b, c549267, a02fdbe, f294652, 3625369, 6dbacb3, and a3fad54		 00:57
kudo/nom: ebc90c1 | (Zoffix Znet)++ | docs/release_guide.pod:
Include example usage of tools/create-release-announcement.pl
 01:21
kudo/nom: e8b2fed | (Zoffix Znet)++ | docs/release_guide.pod:
Use ./perl6 instead of perl6 in the examples

Running perl6 in the built rakudo directory throws mismatch errors
kudo/nom: 83a88b8 | (Zoffix Znet)++ | docs/announce/2016.07.md:
Add 2016.07 release announcement
 01:31
kudo/nom: 06a656a | (Zoffix Znet)++ | docs/release_guide.pod:
List 2016.07 release as now in the past
kudo/nom: e2ec357 | (Zoffix Znet)++ | docs/release_guide.pod:
List Zoffix Znet as release manager for #102 August release

Unless someone else who wants to learn how to do a release wishes to have a go.
travis-ci Rakudo build errored. Zoffix Znet 'Populate ChangeLog 01:49
travis-ci.org/rakudo/rakudo/builds/145152139 github.com/rakudo/rakudo/compare/7...f08d75f69e
Zoffix bisect: 'á' ~~ m:g:ignoremark/ <[a]> / 02:56
bisectable Zoffix: Exit code is 1 on both starting points, bisecting by using the output 02:57
Zoffix: bisect log: gist.github.com/32dffd85b3779b0c63...d2a7f60e12
Zoffix: (2016-03-24) github.com/rakudo/rakudo/commit/f3fe819
AlexDaniel Looking at the log, it seems like the change was in the error message 03:31
Zoffix Yeah 03:32
AlexDaniel committable: f3fe8196 'á' ~~ m:g:ignoremark/ <[a]> /
committable AlexDaniel: |«f3fe819»: ===SORRY!===␤chr codepoint cannot be negative exit code = 1
AlexDaniel committable: f3fe8196~1 'á' ~~ m:g:ignoremark/ <[a]> /
committable AlexDaniel: |«f3fe819»: chr codepoint cannot be negative␤ in block <unit> at /tmp/SeUIEM4fIa line 1␤ exit code = 1
AlexDaniel yeah
MasterDuke benchable: HEAD~50..HEAD my $a = "a" x 2**16;for ^1000 {my $b = $a.chop($_)} 04:13
benchable MasterDuke: gist.github.com/cf753ff1ebad318ed7...7d11751f66 04:15
dalek ast: ad4acf2 | (Zoffix Znet)++ | packages/Test/Util.pm:
Add doesn't-hang() multi that takes Str instead of Capture

Run given code as perl6 -e		 04:25
ast: 1c35ad4 | (Zoffix Znet)++ | S06-signature/slurpy-params.t:
[TODO FUDGE] .gist on @_ containing lazy list correctly thinks it is lazy

RT#128201
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128201
ast: bfd4f0d | usev6++ | S06-signature/slurpy-params.t:
Skip new test on JVM (Proc::Async NYI)
 06:57
p: 3d15ced | peschwa++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/ (2 files):
Fix up ClassLoader chains for jvminterop.
 07:52
kudo/nom: 263b021 | peschwa++ | tools/build/NQP_REVISION:
Bump NQP_REVISION for ClassLoader chaining changes.
kudo/nom: e46407d | peschwa++ | / (7 files):
Bring jvminterop in line with our CUR stuffs.

This now means we have two ways for interop:
  (1) loading classes from the Java Runtime works with :from<JavaRuntime>.
  (2) loading classes from somewhere else on the filesystem works with
  'use lib <java#$path>; use ClassfileOrJarInPath:from<Java>'
Adjust existing tests accordingly.
98d57fb | peschwa++ | t/03-jvm/ (2 files): Add test for "use lib" jvminterop
psch there, that's something \o/ 07:53
bartolin \o/
psch: do you remember the error 'expected Positional but got Seq' on r-j? you mentioned in RT #126493 that it seems to be related to signature compilation 08:08
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126493
bartolin would it make sense to work around that in Binder.java like that: gist.github.com/usev6/aac530b87dd2...b75c7ceaf5
(that's probably a crude patch, but it would make about 40 spectests pass :-) 08:09
psch bartolin: the first inner if seems unnecessary?
bartolin: as in, it's just the same conditional as the previous one, but nested 08:10
bartolin psch: oh, I removed the first conditional (that one with &&)
psch bartolin: ah, okay. yeah, that sounds reasonable to me 08:11
bartolin the first inner if is basically the old code
psch as in, we check "is it nomType gcx.Positional" and then we check the two paths, decont and no-decont
bartolin *nod*
psch well, the thing is that it was a known workaround to decont into the call that throws like that 08:12
and moar does quite a bit more magical {,de}conting i have the feeling :)
so i'd not even call this much of a workaround, but more of a "as good as we can make it at the moment" fix
like, it doesn't actually work around the problem, but simply fixes it in the easiest way currently known vOv
bartolin :-) 08:13
so ... I'll open a PR
psch d'you have anything else off the top of your head that has some amount of tests in double digits?
('cause i don't *really* want to dig into NC right now :P )
bartolin no, not off the top of my head 08:15
psch alright, i'll just grep through roast if i can find anything that looks easy-ish 08:20
i kinda feel like easy-ish bugs atm :P
bartolin ++psch :-) 08:21
FROGGS o/ 08:22
bartolin hi FROGGS!
psch o/ FROGGS
FROGGS :o) 08:23
bartolin psch: PR 827 opened 08:30
psch bartolin: i'll let travis finish (and not go to bed this time, like with i think Zoffix++ PR some time back...) just for formalities sake 08:31
bartolin oh, that PR can wait -- you're welcome to sleep first :-) 08:32
psch it's 10:30 AM... :)
bartolin ahh, same timezone 08:33
FROGGS sure 08:34
same country
bartolin oh! didn't know that 08:35
FROGGS psch: where do you live again? was south-westish Germany, right?
psch FROGGS: no, rather central-westish, actually. Osnabrück currently 08:36
FROGGS ahh, yeah
psch i kinda did have a plan of moving to Berlin, but that's somewhat on hold, due to medical reasons :S 08:37
FROGGS :S 08:38
psch nothing bad though, no worries :)
FROGGS phew :o)
bartolin psch: yes, phew. however, would be very nice to meet you someday (and have a beer together, or something) 08:39
bartolin lives in Berlin
FROGGS almost lives in Berlin
bartolin waves southwards 08:42
FROGGS waves back 08:43
*g*
psch i'm barely 14 geographical minutes south, i don't think that really counts :P 08:44
well, considering it's like 5 degrees west at least
buut, maybe i'm not woven at :P
FROGGS right :P
I'm living 1 thousand meters outside of Berlin, in the south 08:45
bartolin waves westwards, too 08:46
psch ...oh, it's "waved", not "woven", isn't it 08:48
FROGGS I'm not shore
psch anyway, yeah, hi bartolin :P 08:49
hm 08:52
should i commit the .class file for the 'use lib' interop test?
'cause that way we don't have to recompile it
i mean, i'd also commit the .java file so we can add to that for tests if necessary
otoh, i've written the test to compile first now... 08:53
kudo/nom: 1bb86ae | peschwa++ | .gitignore:
Add t/03-jvm/*.class to gitignore
kudo/nom: d09ad6e | usev6++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Look for PositionalBindFailover in decont path as well

This avoids type check errors
  'expected Positional but got Seq'
as described in RT #126493		 09:15
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126493
dalek kudo/nom: cd19db2 | FROGGS++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Merge pull request #827 from usev6/rt126493

Look for PositionalBindFailover in decont path as well
psch ah, the unfinished runs on travis are osx 09:16
FROGGS yeah, I thought the ones that passed are good enough 09:17
psch no objection from me :)
although, again, someone else merged a PR i said i would merge /o\ 09:18
my reputation! D:
FROGGS hehe
psch #124279 is apparently a lot in roast, but that's definitely not easy 09:22
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124279
psch it's actually pretty close to a giant mess spread across 2/3rds of nqp-j /o\
bartolin psch: yeah, I didn't mention that one because you mumbled something like 'easy-ish' :-) 09:25
psch the good thing is, that lots of the remaining trouble we have with CX in general are in that area 09:30
like 'A: for ^1 { B: for ^1 { last A } }', which fails with "CX without handler" or somesuch iirc 09:31
the bad thing is that, well, it's interactions between ExceptionHandling.java and QAST::Compiler that somehow don't do quite the right thing, and i have no real clue where exactly things have to be fixed in what way :|
timotimo i haven't heard them called "environmental variables" yet %) 09:48
dalek ast: c9df7b9 | usev6++ | S (8 files):
Unfudge tests for RT #126493
 09:58
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126493
jnthn afternoon, #perl6-dev 10:22
yoleaux2 15 Jul 2016 06:43Z <Zoffix> jnthn: just a reminder that tomorrow is a Rakudo release and we'll need a MoarVM release cut. Thanks.
jnthn ...that's precisely why I'm here :-)
dalek ast: db4a9a7 | usev6++ | S10-packages/precompilation.t:
Fudge two precomp tests for rakudo-j
bartolin o/ jnthn 10:23
jnthn o/ bartolin 10:25
timotimo o/ bartothn
jnthn can only count Berlin as close by if 4.5 hours of train is "close" :) 10:26
nine jnthn: that's practically in the neighborhood
jnthn Yeah, pretty much. :-)
jnthn likes having both Berlin and Vienna in a bit over 4 hours train ride. :)
If they get around to building the high speed lines they're talking about here, they'll be within 2 hours... 10:27
Wow, quiet MoarVM month. Only 16 commits. 10:29
bartolin would it be possible to re-enable rakudo-j for camelia? or where there other problems apart from 'make install' not working? 10:37
psch i think it'd just need the evalserver started? 10:40
assuming the build is still working, that is
which, actually, might well be disabled...
bartolin doesn't know where that is configured/managed 10:41
jnthn .tell Zoffix www.moarvm.org/releases/MoarVM-2016.07.tar.gz 10:45
yoleaux2 jnthn: I'll pass your message to Zoffix.
nine j: say "Hello world!" 10:58
camelia rakudo-jvm 40a953: OUTPUT«Hello world!␤»
nine j: say "$*PERL"
camelia rakudo-jvm 40a953: OUTPUT«Perl 6␤»
psch j: use java::lang::String:from<JavaRuntime>; say "foo" eqv String.new("foo")
camelia rakudo-jvm 40a953: OUTPUT«===SORRY!===␤Could not find java::lang::String in:␤ /home/camelia␤ /home/camelia/rakudo-j-inst-2/share/perl6/runtime␤ /home/camelia/rakudo-j-inst-2/share/perl6/lib␤ /home/camelia/rakudo-j-inst-2/share/nqp/lib␤ /home/camelia/.perl6…»
nine j: say "$*PERL.compiler" 10:59
camelia rakudo-jvm 40a953: OUTPUT«Perl 6.compiler␤»
psch oohkay that's weird
nine j: say $*PERL.compiler
camelia rakudo-jvm 40a953: OUTPUT«rakudo (2016.03.102.g.40.a.953.f)␤»
nine Looks a tiny bit outdated
psch oh
yeah
j: use java::lang::String:from<Java>; say "foo" eq String.new("foo")
camelia rakudo-jvm 40a953: OUTPUT«===SORRY!===␤Could not find java::lang::String in:␤ /home/camelia␤ /home/camelia/rakudo-j-inst-2/share/perl6/runtime␤ /home/camelia/rakudo-j-inst-2/share/perl6/lib␤ /home/camelia/rakudo-j-inst-2/share/nqp/lib␤ /home/camelia/.perl6…»
psch ya that was still broken there too
nine cronjob for building rakudo-j is still active 11:01
running ~/evalbot/build.pl rakudo-jvm manually now to see how it fails
sh: /home/camelia/rakudo-j-inst-1/bin/nqp-j: No such file or directory 11:04
atal: unable to access 'github.com/perl6/nqp.git/': Failed to connect to github.com port 443: Connection timed out
fatal: unable to access 'github.com/perl6/nqp.git/': Failed to connect to github.com port 443: Connection timed out 11:05
Ok, I officially don't get this. curl github.com/perl6/nqp.git/ works just fine in camelia's VM but git fetch times out. 11:11
psch maybe some git https proxy env vars?
...do those exist? 11:12
nine None that I can find
Also rakudo-m-1/nqp has exactly the same git config 11:13
psch wow, #126528 is a nice bit of crazy 11:33
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=126528
psch result.push(HYPER(&operator, leftv, rightv, :$dwim-left, :$dwim-right)); from metaops.pm:530
that's the line that dies with "can't resolve", because it only sees positional arguments..?
timotimo deeeeeeeeeeeeec1 11:34
bartolin j: say $*PERL.compiler 11:41
camelia rakudo-jvm cd19db: OUTPUT«rakudo (2016.06.276.gcd.19.db.2)␤»
bartolin \o/
nine++
j: my $a = <a b c b d>; $a .= unique; say $a.List 11:43
camelia rakudo-jvm cd19db: OUTPUT«This Seq has already been iterated, and its values consumed␤(you might solve this by adding .cache on usages of the Seq, or␤by assigning the Seq into an array)␤ in block <unit> at <tmp> line 1␤␤»
bartolin j: my $a = <a b c b d>; $a = $a.unique; say $a.List 11:44
camelia rakudo-jvm cd19db: OUTPUT«(a b c d)␤»
nine Seems like sometimes one just has to wait :) 11:45
bartolin found that one earlier today. $a .= unique is supposed to be syntax sugar for $a = $a.unique, isn't it?
nine AFAIK yes
Zoffix jnthn++ thanks 11:47
yoleaux2 10:45Z <jnthn> Zoffix: www.moarvm.org/releases/MoarVM-2016.07.tar.gz
dalek kudo/nom: 88c35ed | (Zoffix Znet)++ | docs/ (2 files):
Reword `is mangled` changes as a bug fix instead of new feature an addition

Per comments github.com/rakudo/rakudo/pull/826#...-233106909		 11:55
Zoffix I can't believe no one thought of naming rakudo-j as rak-judo :) 12:16
psch wouldn't come to my mind. the 'u' is almost silent in the japanese word "rakudo" 12:17
well, depending on dialect probably, but i only have a bit of an idea of anime japanese :P
DrForr It's 'ra ku do', separate syllables. Though there are words like 'handakuten' where the 'u' is almost silent. 12:21
psch ah. who would've thought that popular media is not a comprehensive introduction into the intricacies of a foreign language :/ 12:22
but google translate agrees, so i'm clearly wrong 12:23
DrForr Well, I did have 2 years in college :)
psch so what makes an u almost silent or not? 12:26
'cause i'd guess the kana for handakuten would be ha-n-da-ku-te-n?
which i'd guess would be the same 'ku'
(never mind that it probably wouldn't be written in kana, it's just the only thing i know that expresses phonems) 12:27
nine wwaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaahh 12:28
nine just hates precompilation
psch yes, let's just remove it again. then we also don't have the BEGIN EVAL bug on r-j anymore :P
nine +1
Anyone against?
*** BREAKING NEWS *** Rakudo removes precompilation *** Coffee related shares rise sharply *** 12:29
DrForr Well, as best as I can remember it's because 'da' is a stressed syllable, and further that the 'da' is a 'ha' with a .. well, handakuten mark, which means it's got a stronger accent. 12:40
Syllables like 'da' and 'za' were introduced later than the iroha syllabary, so they get extra accent to make it clearer, at least I think that's the historical background. 12:41
And I just remembered it's 'ta' with the marker, not 'ha'. 12:42
Indirectly related to the fact that native Japanese speakers have trouble with 'ra' and 'la' syllables in English. 12:43
psch that's also regional, isn't it? 12:46
i mean, where between 'ra' and 'la' the syllable is pronounced in japanese
DrForr Japanese doesn't have 'la'. 12:47
Zoffix Is there a way to tell the fudger to skip-fudge the 1 block but manually specify the number of tests it's meant to use in the skip() call? using, e.g. '4 skip' causes it to go hunt for those 4 tests, but I only want it to skip 1 block 12:50
In particular, this fudge isn't working right. It sees 4 tests inside subtest() and uses 4 in skip() call, but there are actually 5 elements in the loop, so I need it to use 5: github.com/perl6/roast/blob/master...code.t#L66 12:51
psch DrForr: en.wikipedia.org/wiki/Japanese_pho...Consonants says the "postalveolar flap" can sound either as 'r' and 'l'..?
s/and/or/ 12:52
ah, en.wikipedia.org/wiki/Dental,_alve...eral_flaps goes into more detail wrt japanese: "[...]actually have a flap that is indeterminate with respect to centrality[...]"
...i don't *really* know linguistics or phonology vOv 12:53
DrForr Yeah, it took me months of practice to get it halfway right. 12:54
nine I've figured that Japanese have neither 'r' nor 'l' but something where the tongue is somewhere in between those positions 12:55
Zoffix found a way to cheat 12:56
DrForr It's just written in ASCII as 'ra' because they needed to choose a letter.
bartolin Zoffix: you can try to insert '#?DOES 20' above the block 12:58
. o O ( or maybe better '#?DOES 5' ) 13:00
Zoffix bartolin, thanks, that works and is cleaner than my hack
Which was #?rakudo.jvm skip 'Only moar handles this', 5); # 13:01
bartolin :-)
dalek ast: 6637413 | (Zoffix Znet)++ | S32-str/encode.t:
Use explicit plan instead of done-testing

As is the case in most other test files		 13:02
ast: 16bcd26 | (Zoffix Znet)++ | S32-str/encode.t:
[TODO FUDGE] .decode.encode roundtrips correctly for utf8-c8

RT#128184
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128184
ast: 58c6aad | (Zoffix Znet)++ | S32-str/encode.t:
Fix incorrect test count in fudge
Zoffix Looks like I'm going to celebrate Rakudo release with 5 steak dinners tonight.... fridge seems to be busted 13:25
jnthn I've no beef with that plan... 13:31
.oO( It's not cool how much food is at steak here... )
 13:32
dalek kudo/nom: b45faa5 | niner++ | src/core/CompUnit/Precompilation (2 files):
Defer creating precomp directories to first write operation

This removes the last remaining write operation from ostensibly pure read code paths.
Fixes permission problems when searching for precomp files in repositories where we lack write permission and reduces clutter.		 13:34
Zoffix :) 13:35
dalek kudo/nom: 0498e80 | niner++ | src/core/CompUnit/PrecompilationRepository.pm:
Enable precompilation even if we change repositories after loading

Replace the restriction by forcing of re-checking of dependencies in cases where there repo chain changed after we already loaded precomp files.		 13:39
timotimo that seems very good 13:42
dalek kudo/nom: b06a99e | niner++ | src/core/CompUnit/PrecompilationRepository.pm:
Fix loading modules from the Staging repo during precompilation

When we change the repo chain after the fact (as in load a precompiled CompUnit::Repository::Staging and then try to load precomp files from there) a new CompUnit::PrecompilationRepository gets created and used. This new precompilation repository would not know about modules loaded by the one used earlier causing precompilation processes to fail.
Fix by sharing the %loaded hash between precompilation repository instances.		 13:46
rakudo/nom: cd42657 | niner++ | / (4 files): 13:50
rakudo/nom: Don't include repositories in precomp dependencies
rakudo/nom:
rakudo/nom: Just because we loaded a custom repo implementation like
rakudo/nom: CompUnit::Repository::Staging during precompilation of a module doesn't
rakudo/nom: mean that the module depends on this repo implementation at runtime.
rakudo/nom:
rakudo/nom: This fixes a bootstrap issue: the Staging repo is for precompiling a
rakudo/nom: module into a staging area which later on gets copied into the final
rakudo/nom: place. The staging area's path however must not end up in the precomp
rakudo/nom: files and should instead be replaced by the repo's logical name if
rakudo/nom: available.
rakudo/nom:
rakudo/nom: When installing CompUnit::Repository::Staging itself however, this
nine Ok, this was the final fix.
jnthn \o/ 13:51
nine++
nine Zoffix: make DESTDIR=... install should work again and we gained a couple of other nice fixes and features in the process.
I'm a bit surprised however that I actually made it through :)
jnthn :) 13:52
Persisting compilation output apparently needs a lot of persistence... :)
Zoffix Weeee \o/ nine++
nine I even left a comment pointing out the place where an interested party could add the one line needed for implementing a global "turn off precompilation" switch ;) 13:59
travis-ci Rakudo build errored. Stefan Seifert 'Defer creating precomp directories to first write operation 14:26
travis-ci.org/rakudo/rakudo/builds/145214203 github.com/rakudo/rakudo/compare/8...5faa533624
dalek p: bd92d41 | (Zoffix Znet)++ | tools/build/MOAR_REVISION:
Bump MoarVM version to 2016.07
 14:29
p: 64e3745 | (Zoffix Znet)++ | VERSION:
bump VERSION to 2016.07
travis-ci Rakudo build failed. Stefan Seifert 'Enable precompilation even if we change repositories after loading 15:03
travis-ci.org/rakudo/rakudo/builds/145214700 github.com/rakudo/rakudo/compare/b...98e803b59c
nine Travis output doesn't look good :/ 15:13
stmuk . o O ( trailing whitespace? )
timotimo nowadays when travis fails i don't even look, because half the time it's just some random problem. github unavailable, travis container hung up near the beginning, ...
Zoffix It's not the last commit. 15:14
timotimo fuck travis 15:15
"this is not an active repository"
yeah right
anyway, Zoffix is right, the latest commit seems to build fine again
the jvm backend ones are still compiling
nine lucky me :) 15:16
Well that's what I get for splitting my changes into individual commits without testing them individually
timotimo the jvm ones seem stuck in the make install phase 15:17
nine Looks like it just takes ages :) 15:19
dalek kudo/nom: c32fc22 | (Zoffix Znet)++ | tools/build/NQP_REVISION:
[release] bump NQP revision
 15:25
kudo/nom: 76ce6d4 | (Zoffix Znet)++ | VERSION:
[release] bump VERSION
 15:27
timotimo goo.gl/photos/YcjeBUgUnV3PSE7SA <- nine, you recently bragged about your pool, eh? :) 15:35
nine Haha :) 15:39
travis-ci Rakudo build errored. Stefan Seifert 'Fix loading modules from the Staging repo during precompilation 15:49
travis-ci.org/rakudo/rakudo/builds/145215487 github.com/rakudo/rakudo/compare/0...6a99ebdd79
Zoffix S11-modules/ require and nested fail on 6.c :( gist.github.com/zoffixznet/6f93685...b268d2efec 16:11
Zoffix starts the run again
"'use lib' may not be pre-compiled" 16:14
gfldex was there an test against the ecosystem with the new release?
Zoffix .ask nine any idea about these failures against 6.c roast? Both fail with use lib may not be precompiled? gist.github.com/zoffixznet/17f266b...2d786e0bbd 16:15
yoleaux2 Zoffix: I'll pass your message to nine.
nine Zoffix: I think those tests are buggy and warrant an -errata commit 16:17
yoleaux2 16:15Z <Zoffix> nine: any idea about these failures against 6.c roast? Both fail with use lib may not be precompiled? gist.github.com/zoffixznet/17f266b...2d786e0bbd
Zoffix Hm... Basically `use lib` inside modules is no longer allowed: gist.github.com/zoffixznet/a60e087...3cbf49ed20 16:19
Well, unless you use no precompilation. 16:20
So anyone with such a setup will have their code broken. Doesn't that make us renege on the promise that stuff passing against 6.c won't break? 16:22
dalek ast/6.c-errata: 32fd216 | niner++ | packages/ (4 files):
Remove useless and actually harmful 'use lib' statements from modules.

There's no need to use lib 't/spec/packages' in a module that _is_ loaded from that path already. If t/spec/packages was not in the repo list, we could not have loaded the modules in the first place.
It's actually harmful as we don't allow a "use lib" in a precompiled module. This only made it into 6.c as other restrictions prevented us from precompiling these modules but those restrictions were an implementation detail, not necessitated by the language design.
Zoffix Cool.
Zoffix builds again
nine Zoffix: the restriction has already been in place. As I explained in the commit message, it was other restrictions that kept us from precompiling.
It's possible that we could actually allow 'use lib' in precopiled modules now, though it scares me and is definitely not last minute material :) 16:23
Zoffix :)
gfldex, the release is still in progress. Ecosystem tests are not part of the process. You can see all steps involved in the release guide: github.com/rakudo/rakudo/blob/nom/..._guide.pod 16:27
gfldex Zoffix: there is a regression in htmlify.p6. Likely in the last 2 days.
i need to make a run to the shop and will dig deeper in 2h or so 16:28
Zoffix gfldex, from what I can see, the failure is in is cached, which is experimental. 16:56
s/in/due/;
Sure takes ages to get enough entropy to make a GPG key :( 17:01
It should use my apartment as source. The place's a mess.
nine Zoffix: run a ping -f on a host on the same network 17:02
Zoffix: also exercise by keeping your mouse moving :)
timotimo twitter.com/DeKroonCat/status/7540...72/video/1 - use cat videos as entropy source
Zoffix 'ping: cannot flood; minimal interval allowed for user is 200ms' 17:03
It's a remote VM. No mouse :)
nine Zoffix: sudo?
Zoffix Oh, right. I have sudo! :)
Oh awesome. That seems to have sped up the process. nine++ 17:04
Zoffix celebrates with an appropriate amount of fun 17:15
</release> 17:16
travis-ci Rakudo build passed. Zoffix Znet '[release] bump NQP revision'
travis-ci.org/rakudo/rakudo/builds/145226597 github.com/rakudo/rakudo/compare/c...2fc2280a71
dalek kudo/nom: a81e028 | (Zoffix Znet)++ | docs/release_guide.pod:
Fix typos in examples filenames
 17:25
kudo/nom: de009f9 | (Zoffix Znet)++ | docs/release_guide.pod:
Use 3-digit template in instructions
kudo/nom: bc722ab | (Zoffix Znet)++ | docs/release_guide.pod:
Add Zoffix as one of the pingables to upload tarballs to rakudo.org
p: 31ca0c2 | (Zoffix Znet)++ | docs/release_guide.pod:
Add Zoffix as one of the pingables to upload tarballs to rakudo.org
 17:26
travis-ci Rakudo build passed. Zoffix Znet '[release] bump VERSION' 17:50
travis-ci.org/rakudo/rakudo/builds/145226787 github.com/rakudo/rakudo/compare/c...ce6d407516
Zoffix m: my %cache; sub foo ($what) { say "wtf"; 42 }; &foo.wrap: -> |c { my $key := c.gist; %cache.EXISTS-KEY( $key ) ?? %cache{ $key } !! (%cache{ $key } := callsame) }; foo(42), foo(55), foo(55); dd %cache 18:00
camelia rakudo-moar bc722a: OUTPUT«wtf␤wtf␤Hash %cache = {"\\(42)" => 42, "\\(55)" => 42}␤»
Zoffix m: my %cache; sub foo ($what) { 42 }; &foo.wrap: -> |c { my $key := c.gist; %cache.EXISTS-KEY( $key ) ?? %cache{ $key } !! (%cache{ $key } := callsame) }; foo(42), foo(55), foo(55); dd %cache
camelia rakudo-moar bc722a: OUTPUT«Hash %cache = {}␤»
Zoffix ¯\_(ツ)_/¯
Can't reproduce the htmlify.p6 failure, but the above is weird.
Oh, it gets optimized away?
m: my %cache; sub foo ($what) { rand }; &foo.wrap: -> |c { my $key := c.gist; %cache.EXISTS-KEY( $key ) ?? %cache{ $key } !! (%cache{ $key } := callsame) }; foo(42), foo(55), foo(55); dd %cache 18:01
camelia rakudo-moar bc722a: OUTPUT«Hash %cache = {"\\(42)" => 0.729925963588579e0, "\\(55)" => 0.408734838943235e0}␤»
Zoffix Seems like it
m: sub cache-it (&sub) { my %cache; &sub.wrap: -> |c { my $key := c.gist; %cache.EXISTS-KEY( $key ) ?? %cache{ $key } !! (%cache{ $key } := callsame) } }; sub foo ($what) { say "meow"; 42 }; cache-it &foo; foo(42), foo(55), foo(55); 18:03
camelia rakudo-moar bc722a: OUTPUT«meow␤meow␤»
Zoffix No idea.
Zoffix blames the default scapegoat for all unknown failures: precompilation :) 18:04
gfldex Zoffix: without is cached the docs build just fine 18:08
a good bit slower tho
Zoffix No idea :/ The above is basically the is cached trait's implementation and it's not failing. 18:09
gfldex in the docs the is cached is also in a module 18:10
Zoffix k, I can reproduce it now 18:23
And hey! I was right, it's precompilation. Disabling it avoids the problem. 18:24
Zoffix will file a ticket after a break
Probably related to this one .wraps across modules: rt.perl.org/Ticket/Display.html?id=128476 18:28
travis-ci Rakudo build errored. Zoffix Znet 'Add Zoffix as one of the pingables to upload tarballs to rakudo.org' 18:35
travis-ci.org/rakudo/rakudo/builds/145240985 github.com/rakudo/rakudo/compare/7...722abe2cd7
nine Zoffix: I'm not really surprised. We can now precompile more than before. 18:42
Zoffix Sweet. 18:52
dalek ast: da3e94a | usev6++ | S0 (5 files):
Unfudge some tests for rakudo-j
 19:36
cygx o/ 19:37
cygx wonders if perl6 should try to jump on the Graal/Truffle bandwagon 19:38
it seems to come along nicely
video at www.youtube.com/watch?v=FJY96_6Y3a4
slides at lafo.ssw.uni-linz.ac.at/pub/papers...ruffle.pdf
benchmark results start on p. 15
Zoffix Isn't that the same thing they were saying about LLVM? "You must do LLVM or else!" 19:45
I can't understand anything the guy in the video is saying :/
cygx Zoffix: it's basically the same thing Parrot was supposed to be (multi-language vm/runtim), but actually competitive
Zoffix neat 19:46
cygx they apparently ended up outperforming any other Ruby implementation, in pathological cases by a factor of 300x 19:47
Zoffix Crap. HTTP::UserAgent install fails now. Last commit was in May, so I assume it's some Rakudo changes that make it fail :/ 20:07
dalek ast: ef7b0da | usev6++ | S04- (2 files):
Unfudge a few skipped tests
 20:09
timotimo if you can figure out a litle golf, we can bisect it :D 20:10
Zoffix bisect: Bool::False // die 20:18
bisectable Zoffix: On both starting points the exit code is 0 and the output is identical as well
Zoffix: Output on both points:
cygx ^-- perhaps that should come with output delimiters 20:20
Zoffix Or maybe I always forced the install due to previously hanging tests and this has nothing to do with Rakudo. I can make the test pass by changing // to || but I don't know if that just masks the failure 20:23
awwaiid that's a nice video on truffle, lots of code
cygx conceptionally, it's not too hard how one should proceed getting Rakudo on Truffle 20:29
1. take 6model/JVM
2. write a MOarvm bytecode interpreter
that would solve all the bootsrapping issues
*bootstrapping
Zoffix bisect: use NativeCall; sub system (Str) is native {}; system 'ls'
bisectable Zoffix: On both starting points the exit code is 0 and the output is identical as well 20:30
Zoffix: Output on both points: appveyor.yml␤blib␤Configure.pl␤CREDITS␤docs␤dynext␤gen␤INSTALL.txt␤lib␤LICENSE␤README.md␤src␤t␤tools␤VERSION
Zoffix bisect: use NativeCall; sub system (Str) is native {}; system 'hostname'
bisectable Zoffix: On both starting points the exit code is 0 and the output is identical as well
Zoffix: Output on both points: superserver
Zoffix This isn't run on hack?
MasterDuke AlexDaniel hasn't moved it their yet 20:31
Zoffix Ah.
MasterDuke *there
how much disk space does hack have? i think timotimo was mentioning something about that recently 20:32
Zoffix 834GB: gist.github.com/zoffixznet/4f6e2ba...a6faa3074a 20:33
MasterDuke ha. as long as the builds go in home i think it'll be fine
timotimo yeah, i'm not worried about space in general, just on / 20:34
which is, as i said, filled up to the brim with logged segfaults or something
MasterDuke pretty sure AlexDaniel said a while ago they are ~70GB
AlexDaniel space is not a problem 20:48
it is about 30MB per commit 20:49
m: say 30*2000
camelia rakudo-moar bc722a: OUTPUT«60000␤»
dalek p: ad2149c | (David Warring)++ | t/nqp/059-nqpop.t:
test nqp::sha1
 20:50
AlexDaniel I am more concerned about CPU usage… :/ 20:51
mst apply 'nice' if you're showing up too much in top? 20:55
AlexDaniel mst: I the most convenient solution for everyone would be to build everything on my server and then rsync it to hack :) 20:59
I think*
dalek p: 1bca321 | (David Warring)++ | t/nqp/059-nqpop.t:
combine new and existing sha1 test
 21:04
lizmat Zoffix++ # release! 21:44
timotimo hooray! 21:45
Zoffix \o/ 21:51
nine Zoffix++ 21:55
dalek kudo/nom: fd1be32 | lizmat++ | src/core/ (2 files):
Make Str.index about 30% faster

By giving it a dedicated Str:D candidate, and lettin the Cool candidate refer to it		 22:24
[TuxCM] This is Rakudo version 2016.07-3-gbc722ab built on MoarVM version 2016.07 22:31
test 16.649
test-t 8.648
csv-parser 17.874
lizmat not too shabby :-) 22:33
dalek kudo/nom: fcfe3c0 | lizmat++ | src/core/ (2 files):
Make Str.rindex about 30% faster

By giving it a dedicated Str:D candidate, and letting the Cool candidate refer to it		 22:35
lizmat good night, #perl6-dev 22:37
timotimo gnite liz 22:38
Zoffix night 22:40
dalek kudo/nom: 7463081 | (Zoffix Znet)++ | docs/ChangeLog:
Start 2016.08 ChangeLog section
 23:14
kudo/nom: 980f496 | (Zoffix Znet)++ | / (3 files):
.polymod with a lazy list must not lose mods if the list runs out

Fixes RT#128176
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128176
ast: 61160f5 | (Zoffix Znet)++ | S32-num/polymod.t:
.polymod with a lazy list does not lose divisors when list runs out

RT#128176		 23:15
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128176
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Make Str.index about 30% faster 23:19
travis-ci.org/rakudo/rakudo/builds/145276565 github.com/rakudo/rakudo/compare/b...1be327fe52
Zoffix m: say 'SS' ~~ m:i/ß/ 23:41
camelia rakudo-moar 980f49: OUTPUT«False␤»
Zoffix m: say 'SS'.fc ~~ 'ß'.fc
camelia rakudo-moar 980f49: OUTPUT«True␤»
Zoffix m: say 'ß' ~~ m:i/SS/ 23:44
camelia rakudo-moar 980f49: OUTPUT«False␤»
gfldex m: <abß deö>
camelia rakudo-moar 980f49: OUTPUT«WARNINGS for <tmp>:␤Useless use of constant value abß deö in sink context (lines 1, 1, 1, 1)␤»
Zoffix Ticketed: <Zoffix> m: say 'ß' ~~ m:i/SS/ 23:45
<camelia> rakudo-moar 980f49: OUTPUT«False␤»
oops
rt.perl.org/Ticket/Display.html?id=128642
gfldex Zoffix: i don't agree on that one. Not all SS mean upper case for 'ß'. 23:46
Zoffix: or "Waffen SS" ~~ "Waffen ß" would be true, making processing of german histroy books quite tricky 23:47
Zoffix m: "Waffen SS".fc eq "Waffen ß".fc 23:48
camelia rakudo-moar 980f49: OUTPUT«WARNINGS for <tmp>:␤Useless use of "eq" in expression ".fc eq \"Waffen ß\".fc" in sink context (line 1)␤»
Zoffix m: say "Waffen SS".fc eq "Waffen ß".fc
camelia rakudo-moar 980f49: OUTPUT«True␤»
Zoffix I don't see a difference between using case-insensitive `eq` above vs case-insensitive regex match. But they give different results. 23:49
gfldex that's why I was told in school that upper case for ß is SZ. Years later my officer agreed.
Zoffix m: say "Waffen SS" eq "Waffen ß".fc 23:51
camelia rakudo-moar 980f49: OUTPUT«False␤»
Zoffix m: say "Waffen SS".fc eq "Waffen ß".fc
camelia rakudo-moar 980f49: OUTPUT«True␤»
Zoffix m: say 'ß'.fc ~~ m:i/SS/
camelia rakudo-moar 980f49: OUTPUT«｢ss｣␤»
Zoffix Ahh

Powered by Cro and the Raku® Programming Language.

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

Thank you!