MasterDuke m: my @a = [[1..3], [4..6], [7..9]]; @a[*-1]:delete; @a.say; @a[0]:delete; @a.say; 00:09
camelia rakudo-moar 58cf9d: OUTPUT«[[1 2 3] [4 5 6]]␤[(Any) [4 5 6]]␤»
MasterDuke is ^^^ expected? :deleting off the end truncates the Array, :deleting off the front leaves a hole
timotimo i think so, yeah 00:10
MasterDuke i have to agree with tushar over in #perl6 that's not what i would have expected 00:16
i can see all sort of bugs coming from people programmatically :deleting stuff that happens to fall at one end or the other and getting different results 00:18
timotimo accessing it via index won't asplode at least
when using Z metaops, that can cause problems, as that stops at the shorter sequence 00:19
Zoffix New post: ""Perl 6: What Programming In The Future Is Like?" (Lightning Talk Slides and Video)": 02:16
dalek kudo/nom: 415461b | (Zoffix Znet)++ | src/core/
Remove now-useless .defined check

Pointed out by b2gills++
The addition of the :D constraint[^1] in the signature already ensures the value is always defined, so we do not need to check that.
llfourn psch: # this might be relevant to your interests in case you didn't see it 02:45
travis-ci Rakudo build failed. Zoffix Znet 'Remove now-useless .defined check 03:36
buggable [travis build above] ☠ Did not recognize some failures. Check results manually 03:37
Zoffix t/04-nativecall/13-union.t 03:38
MasterDuke if you run this in your rakudo directory: perl6 -e 'for run(:out, "git", ‘log’, ‘-z’, ‘--pretty=%H’, ‘2015.07^..HEAD’).out.split(0.chr, :skip-empty) { last }' 04:37
you get this error: fatal: write failure on 'stdout': Connection reset by peer 04:38
if you take out that 'last' in the for body (or replace it with just about anything else), no error 04:39
is that unavoidable, or is there a way for rakudo to more gracefully exit the loop (and therefore the run)? 04:41
geekosaur the convention is for programs to ignore SIGPIPE specifically to avoid erroring in that situation. 04:49
if git does not do so, you cannot force it to do so
er, not ignore, but not issue errors on it.
dalek ast: 0806465 | usev6++ | S32-num/int.t:
Fudge newly added test for JVM
geekosaur well, unless we do something like leave SIGPIPE at SIG_IGN in the spawned child 05:10
psch llfourn: i hadn't seen it, thanks. the idea sounds fun 07:41
i wonder if we already can simply create a CompUnit for P6Regex instead of perl6...
also, what's the moar equivalent of jvms Ops.typeName? do we have that? 07:45
humm, how do i reply to a given mailing list topic without having recieved it because i wasn't subscribed yet 08:25
DrForr Have someone forward it to you? 08:29
psch hm, the list's help says i can request specific messages 08:32
but that'd mean i'd have to know how many there were already afaict..?
DrForr Is there a mailman-ish interface? 08:33
psch it's apparently ezmlm
so, uhm, i don't know. i don't know what the mailman interface looks like :) 08:35
stmuk_ psch: add the right header 08:36
psch ah, apparently the index command clips to the latest messages if the argument is too large 08:39
and with that i can get the full messages now
stmuk_: i have no idea how i'd figure out the header, or even how to set it in gmail :S 08:42
also, yes, i did get the full PCRE6/JVM thread, but it's still a digest from [email@hidden.address] which probably means i can't reply like this...
stmuk_ you could set up mutt to use google imap .. I think it's something like x-ref: 09:19
or not bother 09:20
psch yeah, probably going with that last one in the end :) 09:21
dalek p: f4e333a | peschwa++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/
Die on CStruct without any fields.
kudo/nom: 7e98695 | peschwa++ | tools/build/NQP_REVISION:
Bump NQP_REVISION for illegal fieldless CStruct
kudo/nom: 5a43547 | peschwa++ | t/04-nativecall/16-rt125408.t:
Fix test for rt125408.

The repr doesn't actually play a role in the original bug, as per
kudo/nom: acd185a | peschwa++ | t/04-nativecall/06-struct.t:
Add a test for invalidity of empty CStruct

That is, classes with repr<CStruct> cannot have no Attributes.
psch hm, the .travis.yml change got lot in dalek it seems 09:38
anyway, travis will probably complain about moar failing a lot, which is the new CStruct test, which needs
plus bumping as approriate
(where 'a lot' means 'on all platform/version combinations there'll be at least one test failure') 09:39
...or something like that
Zoffix j: Int ~~ 42 10:28
camelia rakudo-jvm 2a1605: OUTPUT«Invocant requires an instance, but a type object was passed␤ in block <unit> at <tmp> line 1␤␤»
Zoffix m: Int ~~ 42
camelia ( no output )
Zoffix ah parsing error
r: Int ~~ τ 10:30
camelia ( no output )
..rakudo-jvm 2a1605: OUTPUT«Invocant requires an instance, but a type object was passed␤ in block <unit> at <tmp> line 1␤␤»
Zoffix Why would there be "parsing errors" with JVM? 10:31
j: say τ
camelia rakudo-jvm 2a1605: OUTPUT«6.283185307179586␤»
Zoffix j: for 42, τ, ½ -> $what { say $what }
camelia rakudo-jvm 2a1605: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Missing block␤at <tmp>:1␤------> for 42, τ,⏏ ½ -> $what { say $what }␤ expecting any of:␤ block or pointy block␤ infix␤ infix stopper␤ prefix␤ …»
Zoffix j: for (42, τ, ½) -> $what { say $what }
camelia rakudo-jvm 2a1605: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Bogus term␤at <tmp>:1␤------> for (42, τ,⏏ ½) -> $what { say $what }␤ expecting any of:␤ infix␤ infix stopper␤ prefix␤ statement end␤ statement…»
Zoffix 0.o 10:32
jnthn Unicode property oddities maybe?
Zoffix Ah
jnthn j: for 42, τ -> $what { say $what }
camelia rakudo-jvm 2a1605: OUTPUT«42␤6.283185307179586␤»
jnthn j: say ½
camelia rakudo-jvm 2a1605: OUTPUT«===SORRY!===␤Argument to "say" seems to be malformed␤at <tmp>:1␤------> say⏏ ½␤Bogus postfix␤at <tmp>:1␤------> say ⏏½␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statemen…»
jnthn m: say ½
camelia rakudo-moar 83d733: OUTPUT«0.5␤»
jnthn Seems it doesn't handle ½ properly 10:33
.u ½
jnthn m: say uniname '½'
camelia rakudo-moar 83d733: OUTPUT«VULGAR FRACTION ONE HALF␤»
jnthn j: say uniname '½'
camelia rakudo-jvm 2a1605: OUTPUT«VULGAR FRACTION ONE HALF␤»
jnthn Hm, it knows it exists at least 10:34
psch m: say uniprop '½' 10:36
camelia rakudo-moar 83d733: OUTPUT«No␤»
psch j: say uniprop '½'
camelia rakudo-jvm 2a1605: OUTPUT«uniprop NYI on jvm backend␤ in block <unit> at <tmp> line 1␤␤»
psch oh, NYI i think?
dalek ast: afbc6b0 | (Zoffix Znet)++ | S32-num/int.t:
Revert "Fudge newly added test for JVM"

This reverts commit 08064659b7897a623daf9f4ee2ba511de131bae8.
The fudge is needed not due to test itself but due to components used in it. The test will be changed to use working compontents in next commit.
ast: 74d1a70 | (Zoffix Znet)++ | S32-num/int.t:
Don't use ½ as it's not yet supported on JVM
psch meh, marshalling troubles :l 10:49
as in, can't get the category of '½' via interop because Character only deals with char and int, not long 10:50
aha, but Integer doesn't get type wrangled... 10:52
j: use java::lang::Character:from<JavaRuntime>; use java::lang::Integer:from<JavaRuntime>; say Character.getType(Integer.valueOf(0xbd))
camelia rakudo-jvm 2a1605: OUTPUT«11␤»
psch m: say '½'.NFD
camelia rakudo-moar 83d733: OUTPUT«NFD:0x<00bd>␤»
psch so, yeah, jvm also thinks '½' is No -- that's the name of the const that is 11 :P
but iirc we do some category handling ourselves, for some reason
oh, and travis did react exactly as i anticipated 10:53
but we had a failure before already so there's no announcement..?
jnthn It'd probably have been better to push the test update after the Moar thing got merged :P 11:05
psch dunno, nine said it'd be fine :) 11:06
jnthn #dammitnine :P 11:09
hackedNO1 NeuralAnomaly: status 12:51
NeuralAnomaly hackedNO1, [✘] Next release will be in 1 day and 2 weeks. Since last release, there are 24 new still-open tickets (21 unreviewed and 0 blockers) and 93 unreviewed commits. See for details
hackedNO1 2 weeks.. time flies
jnthn o.O 12:52
[Coke] finds "1 day and 2 weeks" non-english. 13:11
[Coke] wonders the origin of that phrasing is. 13:13
jnthn It's English, just odd :P 13:16
otoh, so is writing dates mm-dd-yyyy :P
hackedNO1 It's an apparent bug in my Perl 5 module :) 13:18
jnthn As usual, language is full of nice inconsistencies :)
hackedNO1 Supposed to say 2 weeks and 1 day
So the origin of that phrase is poor hacking skills :D 13:19
jnthn :)
perlpilot it should clearly say "1 fortnight and 1 day" ;) 13:20
jnthn "far enough away you can still chill" 13:21
|Tux| Better late than not ... 13:37
This is Rakudo version 2016.09-93-g83d733c built on MThis is Rakudo version 2016.09-93-g83d733c built on MoarVM version 2016.09-3-g9b39aa5 13:38
csv-ip5xs 4.237
test 15.842
test-t 7.066
csv-parser 17.689
oarVM version 2016.09-3-g9b39aa5
csv-ip5xs 4.237
test 15.842
test-t 7.066
csv-parser 17.689
timotimo pretty good performance there
RabidGravy perl 5 might be faster at somethings than rakudo but it takes twice as long to build on this machine 14:07
arnsholt Which numbers are what in those CSV numbers? 14:08
hackedNO1 Isn't that just the tets?
timotimo ip5xs is inline::perl5 for the XS version of Text::CSV
hackedNO1 Try running the stresstest along with the build on that box :P
timotimo if i remember correctly, test is a super naive implementation of csv, and test-t is the optimized, smart implementation? 14:09
hackedNODE Hoe come slurpies can't have defaults? 20:27
gfldex jnthn: it seams I got a Channel without a start that tries to .rethrow and hangs (likely in .rethrow). Could that be? 20:29
jnthn: not related to no start. Channel and .rethrow seams to cause a hang. 20:31
jnthn doesn't see the connection between channels and rethrow... 20:59
Got a golfed example? 21:04
gfldex jnthn: yes: 21:53
dalek kudo/nom: 31eba77 | usev6++ | src/vm/jvm/runtime/org/perl6/rakudo/
Report type name in type constraint error on JVM

If an instance object was passed where a type object was required, or vice versa, Rakudo on MoarVM already reported the name of the required type (commit 9a1616ed63).
Now Rakudo on JVM does the same. Fixes RT #126124 for rakudo-j.
synopsebot6 Link:
dalek rakudo/nom: 1f29cb2 | lizmat++ | src/vm/jvm/runtime/org/perl6/rakudo/
rakudo/nom: Merge pull request #891 from usev6/rt126124
gfldex jnthn: shall I rakudobug? (i'm not quite sure if it's actually channel related) 21:57
MasterDuke jnthn: btw, if you're interested, i seem to have a different segfault's valgrind output here 21:58
timotimo gfldex: isn't that just because you're never closing the channel?
and .list might be trying to .eager it?
oh, say is definitely trying to .eager it
gfldex if you comment the die out, it works 21:59
timotimo of course
because then the $channel.close will be executed
gfldex i will try to LEAVE the close
timotimo good idea
gfldex that solves the problem 22:00
jnthn: sorry, ENOBUG 22:01
timotimo :)
glad it was so easy
gfldex was there a decision made what happens to exceptions thrown inside a thread (that is not the main thread), that are not caught by an await? 22:06
timotimo no, it should break the promise 22:07
gfldex the promise is kinda list in my example
timotimo right, you're not awaiting that promise 22:08
line 124-138 in case you wonder
timotimo we wanted to have something like warning if you .sink a promise 22:09
because then if it gets broken nothing notices that
you know there's something like .then that you can use to close the channel if the promise gets broken? 22:11
gfldex i have a CATCH block in this example so a LEAVE seams fitting 22:16
timotimo yup 22:17
lizmat good night, #perl6-dev 22:18
timotimo gnite lizmat!
From: "Carl Mäsak" (via RT) [email@hidden.address] 22:20
masak: how could you? :) 22:21
geekosaur double encoding rears its head again 22:23
(aka wtf8)
timotimo wtf88 22:24
geekosaur hm, no, that;s the other problem. utf8 being "interpreted" by iso8859
timotimo ah, ok
well, headers in mails are supposed to only ever be iso8859, or something like that?
geekosaur which is more likely to be the receiving mail server's fault
timotimo i expect my client caused the b0rkage at the end 22:25
geekosaur only supposed to be 0x20-0x7f although there is an extension
timotimo X-Amavis-Alert: BAD HEADER SECTION, Non-encoded 8-bit data (char C3 hex):
From: "Carl M\303\203\302\244sak" (via RT) [...]
yeah, i think there's a way to put something in front of a header's value to say "oh btw this is encoded in ***"
geekosaur ... *that* looks double encoded
timotimo m: say 0xc3.base(8) 22:26
camelia rakudo-moar 1f29cb: OUTPUT«303␤»
jnthn gfldex: Sorry, was away playing TTR. Glad I don't have a bug to hunt this time. ;) 22:27
MasterDuke: That's likely harmless, though a bit sloppy. (I suspect it's just that we pad segments in the bytecode file, and never care about what value the padding has.) 22:28
MasterDuke: Feel free to drop that in the MoarVM queue. You can even hacktoberfest it. ;)
timotimo Transport Tycoon Reloaded? 22:29
gfldex is there any propaganda about hacktoberfest? 22:30
timotimo pandaganda 22:31
AlexDaniel jnthn: how can segfault be harmless?
jnthn AlexDaniel: Where was there a segfault? 22:33
gfldex i wonder if IO::Path should does Associative
jnthn Oh, odd
The gist is marked segfault
gfldex '/home/'<dex>:exists
jnthn But the valgrind output doesn't indicate one
MasterDuke *** Error in `/home/bisectable/.rakudobrew/moar-nom/install/bin/moar': double free or corruption (fasttop): 0x00000000047e7ea0 *** Aborted 22:34
jnthn That's not in the gist?
MasterDuke no, that's when running not under valgrind
jnthn 'fraid those two are unrelated. 22:35
MasterDuke running the same script in a loop
i'll keep running it in a loop with valgrind then, hopefully get a segfault
jnthn Yeah, valgrind should output SIGSEGV, or at the very least something about an invalid read or invalid write 22:36
It'll reliably catch and warn about the bytecode padding thingy too...which is also worth filing a MoarVM issue on. 22:37
Especially if you've a reliable reproduction
MasterDuke 22:38
jnthn Thanks
I was hunting a SEGV earlier this week (didn't find it yet) that only occasionally showed up under valgrind... is what I managed to catch, fwiw 22:40
MasterDuke so the backstory is that the script that builds new commits for the whateverables wiped out the directory it runs from (deleting almost everything else in the directory) 22:43
jnthn o.O
MasterDuke and we're trying to figure out what happened/where in the code 22:44
AlexDaniel MasterDuke: that's an older version I think
MasterDuke 22:45
current version
jnthn fwiw, that poll loop thingy could be start react { whenever $channel -> $commit { ... } } 22:47
Well, provided a .close happens at some point 22:48
Though .race will be nicer :)
jnthn wonders if it could be something about LEAVE phaser scoping issues... :S 22:49
AlexDaniel jnthn: there are many issues there :( For example, 「sleep rand / 10;」 is *very* required 22:50
travis-ci Rakudo build failed. lizmat 'Merge pull request #891 from usev6/rt126124
AlexDaniel but it takes time to isolate all of the actual bugs
buggable [travis build above] ☠ Did not recognize some failures. Check results manually 22:52
jnthn AlexDaniel: Debugging/fixing them is more than a little time-consuming too
AlexDaniel jnthn: yeah I know
jnthn Ah well, torture the implementors... :) 22:55
jnthn gets some rest... 'night 22:58
MasterDuke wonders if it would be bad to wish jnthn a night of finding segfault solutions in his dreams 23:02
japhb .ask timotimo Have you had a chance to work on again? I'm curious how `method enqueue(int $color, int $x, int $y)` turned out .... 23:06
yoleaux2 japhb: I'll pass your message to timotimo.
timotimo o/ 23:30
yoleaux2 23:06Z <japhb> timotimo: Have you had a chance to work on again? I'm curious how `method enqueue(int $color, int $x, int $y)` turned out ....
timotimo japhb: i think i tried it and it was almost useless :)
yeah, in fact i have it open in front of me