Zoffix F's sake >_< spent an X-amount of time trying to debug why my change was breaking a test and turns out the breakage was due to my editor auto-removing trailing spaces ~_~ 00:36
btyler_ timotimo: well, I gave a talk about erlang and perl6 and schedulers :) 00:58
in #perl6? will backlog 00:59
dalek ast: 2d2efaa | (Zoffix Znet)++ | S26-documentation/07-tables.t:
Add a warning for trailing-whitespace tests

It's not uncommon to configure editors to auto-strip trailing whitespace and this feature silently breaks tests in this file.		 01:22
ast: a7387a6 | (Zoffix Znet)++ | S26-documentation/07-tables.t:
Merge pull request #129 from zoffixznet/fix-RT128221-fail-table-parse-leading-hyphen

Add a warning for trailing-whitespace tests
kudo/nom: 57f8f8f | lizmat++ | src/core/Str.pm:
Fix for RT #128034

Inspired by PR #796, Zoffix++		 03:10
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128034
[Tux] This is Rakudo version 2016.06-9-g57f8f8f built on MoarVM version 2016.06 06:20
test 16.529
test-t 9.965
csv-parser 21.823
next run was 9.86 06:26
Zoffix lizmat++ good fix 10:53
masak is it really the *sanity* we're limiting...? 12:14
you're sure it's not (say) the insanity? :P 12:15
BrokenRobot :)
masak likely didn't get-the-joke in that commit, or it's an honest overnegation
BrokenRobot it's a sanity check for `limit' 12:16
[Coke] (mailed) Not that this helps, but I think fax is also acceptable. Not sure if scan&email is, but probably? 12:19
if the negation kind of gets away from you, is that an abnegation? 12:20
(because ab-, ... nevermind)
lizmat backlogs 12:24
dalek ast: c3f40c4 | (Zoffix Znet)++ | S32-str/split-simple.t:
Add tests for RT#128034 (split with NaN limit must throw)
 12:28
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128034
masak [Coke]: ;) 12:31
BrokenRobot: ah, that explains why I read it wrong. I'm used to routine names being verbs when they can. 12:32
lizmat masak: limit-sanity seemed like a good dual of split-sanity :-) 12:41
masak now I know to read those as `ensure-limit-sanity` and `ensure-split-sanity` ;) 12:46
[Coke] thanks to hoelzro++ and Zoffix++ for doing the last 2 releases@ 12:48
!
BrokenRobot \o/ 12:51
dalek kudo/nom: 7cff429 | lizmat++ | src/core/Str.pm:
Take masak++ 's implicit naming suggestion
 12:52
[Coke] m: @// '_'; # perl6 emojis 12:55
camelia ( no output )
lizmat wonders what caused the test-t regression
BrokenRobot split NaN test? There only three other commits that went in since last 9.46 for test-t: github.com/rakudo/rakudo/commit/a223163b3c (is required with benefits) github.com/rakudo/rakudo/commit/8ad19e2ec3 (Backtrace::Frame own new) github.com/rakudo/rakudo/commit/ee663c40ac (Don't leak numeric cases) 13:07
BrokenRobot imagines a CSV module would heavily use .split
BrokenRobot wonders how representative the test-t stuff as a bench for the entire fastness of rakudo 13:08
lizmat BrokenRobot: well, it is *a* canary of some sort 13:29
definitely not *the* canary :-)
dalek kudo/nom: 2c61fda | peschwa++ | src/core/StrDistance.pm:
Make StrDistance.Str return $!after.
 13:45
ast/6.c-errata: 45aede9 | peschwa++ | S05-transliteration/trans.t:
Add test for stringification of StrDistance
psch oh fsck 13:46
was that the right branch?
'cause i think it wasn't
BrokenRobot master, I think?
Well, that's where I was adding things at least :P
dalek ast: 20103df | peschwa++ | S05-transliteration/trans.t:
Add test for stringification of StrDistance
 13:47
psch i'll revert on 6.c-errata i suppose 13:48
dalek ast/6.c-errata: 8367c1a | peschwa++ | S05-transliteration/trans.t:
Revert "Add test for stringification of StrDistance"

This reverts commit 45aede9c07ceaf876eec04a2607a222b8c623564. Wrongfully commited on 6.c-errata, hence it goes away again, sorry.
BrokenRobot m: class Foo { method check { for ^1000_000 { next unless 5 != $_ } } }.new.check; say now - INIT now 13:52
camelia rakudo-moar 2c61fd: OUTPUTĀ«0.49098248ā¤Ā»
BrokenRobot m: class Foo { method !x (\a, \b) { a != b }; method check { for ^1000_000 { next unless self!x(5, $_) } } }.new.check; say now - INIT now
camelia rakudo-moar 2c61fd: OUTPUTĀ«0.6636506ā¤Ā»
BrokenRobot m: say (0.6636506 - 0.49098248) / 0.6636506
camelia rakudo-moar 2c61fd: OUTPUTĀ«0.260179257ā¤Ā»
BrokenRobot lizmat: an extra method call is 26% slower. If CSV uses a ton of splits maybe it could be *the* canary? 13:54
lizmat yes, CSV does a lot of split, but in a profile the !limit-sanity was at 2% CPU, and it replaces some other tests 13:56
so, I'm not sure what's going on
no time to dig deeper atm, busy with P6W
BrokenRobot Ah. ok. :) I need to learn more about how to do performance tests. 14:12
lizmat And another Perl 6 weekly hits the Net: p6weekly.wordpress.com/2016/06/21/...m-orlando/ 14:55
nine Thanks lizmat :) 14:58
lizmat m: class A { has $.a }; A.new(a=>42).a = 666 # I think this is actually an LTA error message 15:31
camelia rakudo-moar 2c61fd: OUTPUTĀ«Cannot modify an immutable Intā¤ in block <unit> at <tmp> line 1ā¤ā¤Ā»
jnthn What do you want it to say? "Cannot assign to"?
lizmat it should probably mention that you were trying to assign to an attribute ? 15:33
a ro attribute ?
jnthn Oh
No, we don't have the info handy to do that
And can't easily have that either
All we have by the time = runs is, well, that immutable Int
lizmat yeah, I realize that 15:34
jnthn So we could word-smith it to mention assignmnet, but I don't know we can do much better
Or at least, I don't see how we can 15:35
lizmat yeah, I have some half-baked ideas... but maybe for much later :[)
jnthn I guess there may be some way to stick some kind of context annotation (in the same way we do file/line numbers) in there and be able to figure it out from the stack trace. 15:37
[Coke] question on the is required updates - was there a ticket for that? or discussion on channel? 15:48
I don't think there was a design doc on it, since it was added as a result of an Ovid slide deck, originally) 15:49
dalek kudo/nom: 5801810 | lizmat++ | t/04-nativecall/13-union.t:
s/is_approx/is-approx/

This appears to fix test errors observe by dha++		 18:02
lizmat argh 18:03
dalek kudo/nom: 479628e | lizmat++ | t/04-nativecall/ (7 files):
s/is_approx/is-approx/ for all nativecall tests

This should remove deprecated warnings		 18:12
psch i've taken another unbiased look at the jvm precomp issue, and i think i went down the wrong track somewhere earlier 18:39
as in, the missing CodeRef is inside the module that was precompiled
the EVAL that happens in the CU that wants to load that module might or might not play a role directly
but the degenerate case of < module Foo { BEGIN EVAL 'say "hi"' } > dies due to a missing CodeRef in the classfile for Foo 18:40
but it does so *after* the BEGIN code ran
so i am currently suspecting that we lose the mainline of the module somehow, and not the nested CU
i still haven't found a reliable way to map existing CodeRefs in a classfile to "pieces of code", so to speak 18:41
i mean, that module Foo up there has 11 CodeRefs
i see maybe three, from where i'm standing, so i'm clearly missing something
lizmat hmmm... I just found out that at least Map.WHICH and BagHash.WHICH are borked 20:00
brb
dalek kudo/nom: 1bf7270 | lizmat++ | src/core/CompUnit/PrecompilationStore/File.pm:
Make sure BUILD is a submethod and returns Nil
 21:02
kudo/nom: d9956e5 | lizmat++ | src/core/ (2 files):
Make Map a value object
kudo/nom: 2058a92 | lizmat++ | src/core/Hash.pm:
Oops, forgot a signature
 21:13
kudo/nom: 5ca24cc | lizmat++ | src/core/ (2 files):
BagHash|MixHash shouldn't be value types
lizmat afk& 21:46

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!