Geth rakudo: FCO++ created pull request #1029:
add ⁇ ‼ unicode version
roast: FCO++ created pull request #246:
Add tests for unicode version of ?? !!
travis-ci Rakudo build errored. Zoffix Znet 'Test Nil line return in REPL ends up as Nil in output 04:59
moritz IOninja: you now have direct access to star 06:39
[Tux] In the original script I now see: 07:25
Tue 14-03-201720:00HAWADe Burchttwee maal verplaatst: een keer naar voren, en een keer terug
perl6 test P6opaque: no such attribute '$storage' in type Map when trying to bind a value
(ignore the middle line :)
This is Rakudo version 2017.02-133-g7966dad5e built on MoarVM version 2017.02-20-g773711e1 07:27
csv-ip5xs 2.809
test-t 4.973 - 5.104
csv-parser 12.822
lizmat Files=1177, Tests=55954, 190 wallclock secs (11.43 usr 4.51 sys + 1131.51 cusr 106.98 csys = 1254.43 CPU) 09:59
IOninja moritz++ thank 11:00
[TuxCM]: is that good or bad? 11:01
Or the error is what you see.
IOninja is mildly amused #1029 fails 25 stresstest files before just hanging the run. 11:32
samcv: and yeah, the [ || ... ] and [ | ... ] is used all over the place. Lets you have an nice straight line of "||" before each conditional 11:35
samcv that's fine with me then :) 11:36
jnthn It's parsed "specially" so the leading one is a no-op 11:39
samcv night all 11:40
IOninja night
[TuxCM] IOninja, it is irrelevant: it is the old initial script. I'll have a look later to see if that can be fixed 11:48
IOninja [TuxCM]: which one is the "original script"? ``? 11:49
[TuxCM] yep
IOninja ./lib/Text/ multi method column_names (False) 11:51
That method won't ever be called BTW. You possibly want (Bool:D where {False})
Or not 11:52
[TuxCM] that script was the first functional draft, so it will not ever be production code
IOninja m: multi x($) { say "here" }; multi x(Bool:D where {False}) {say "there"}; x False
camelia 5===SORRY!5=== Error while compiling <tmp>
Cannot do non-typename cases of type_constraint yet
at <tmp>:1
------> 3y "here" }; multi x(Bool:D where {False}7⏏5) {say "there"}; x False
IOninja m: multi x($) { say "here" }; multi x(Bool:D $ where {False}) {say "there"}; x False
camelia here
IOninja m: multi x($) { say "here" }; multi x(Bool:D $ where *.not) {say "there"}; x False
camelia there
IOninja Ah. OK then :)
[TuxCM] «my $line = callframe(1).annotations<line>;» is what is causing the fail 11:54
IOninja Text::CSV is in the repo tho and multi method column_names (False) is in it 11:55
m: callframe(1).annotations<line>
camelia P6opaque: no such attribute '$storage' in type Map when trying to bind a value
in block <unit> at <tmp> line 1
IOninja Thanks.
bisect: callframe(1).annotations<line>
bisectable6 IOninja, Bisecting by exit code (old=2015.12 new=7966dad). Old exit code: 0
lizmat jnthn: so I have a patch that will make .reverse return a Seq, it needs some spectest tweaking, for a 10 element array it is about 30% faster, for a 1000 element array about the same, but reduces the number of GC's to 1/3 of the original 11:56
IOninja Ah, see it. 11:57
Typo on this line: 11:58
'$storage' neds to be '$!storage'
I'd fix it, but gonna run to work./
lizmat jnthn: do we consider this worth the potential disruption (my @a := @b.reverse potentially breaking)
IOninja: oops, will fix
jnthn lizmat: What did you test? Iterating over it? 11:59
[TuxCM] (canary in the coalmine)++
jnthn (to get the 30% I mean)
Geth rakudo/nom: 9a74cd0e51 | (Elizabeth Mattijsen)++ | src/core/
Fix typo, spotted by Zoffix++ Tux++
lizmat my @a = ^10; for ^100000 { for @a.reverse { } } # jnthn 12:00
brb 12:03
Geth rakudo/nom: beda5767ba | (Elizabeth Mattijsen)++ | 4 files
Make List.reverse return a Seq

This appears to have been the intent in the speculation all along. Reason for doing this is mainly to reduce memory pressure, as the previous implementation would *always* create a copy of the reified list. By making it a Seq, it will only create a copy if that is needed. ... (13 more lines)
roast: 93bd430e1a | (Elizabeth Mattijsen)++ | 4 files
Handle fallout of List.reverse -> Seq migration
lizmat will handle the 6.c-errata case before the release if these commits are not reverted 12:53
afk for a bit longer& 12:54
Geth roast: 749ed327f6 | (Zoffix Znet)++ | S06-advanced/callframe.t
Test callframe.annotations does not crash

Rakudo fix:
rakudo/nom: cf935ca58f | (Zoffix Znet)++ | t/05-messages/01-errors.t
Test Str.Rat.nude on non-numerics doesn't reference guts

synopsebot6 Link:
Geth roast: 9c4b563910 | (Zoffix Znet)++ | S03-metaops/cross.t
Test `X` works with lazy RHS

RT#130566: Rakudo fix:
synopsebot6 Link:
rakudo/nom: c5ab28fe2f | (Zoffix Znet)++ | t/05-messages/01-errors.t
Use same RT ticket labeling format as in roast
rakudo/nom: 53a6d3aff6 | (Zoffix Znet)++ | t/05-messages/01-errors.t
Test stub code does not reference guts in error when executed

synopsebot6 Link:
Geth rakudo/Seq-eqv-List-False: f9eb811dd2 | (Zoffix Znet)++ | src/core/
Make Seq:D eqv List:D return False even when all elements are same
rakudo: zoffixznet++ created pull request #1030:
Make Seq:D eqv List:D return False even when all elements are same
roast: 846e330d96 | (Zoffix Znet)++ | S32-list/rotor.t
Test .rotor does not hang with infinite iterable as the cycle

RT#129175: Rakudo fix:
synopsebot6 Link:
roast: 3fdfcc0ca3 | (Zoffix Znet)++ | S06-signature/slurpy-params.t
Test slurpy positional param does not hang when given inf. list

RT#129175: Rakudo fix:
synopsebot6 Link:
rakudo/nom: f9eb811dd2 | (Zoffix Znet)++ | src/core/
Make Seq:D eqv List:D return False even when all elements are same
rakudo/nom: dd1dab2e35 | (Zoffix Znet)++ | src/core/
Merge pull request #1030 from rakudo/Seq-eqv-List-False

Make Seq:D eqv List:D return False even when all elements are same
roast: dcdcec2441 | (Zoffix Znet)++ | S03-operators/eqv.t
Test Seqs !eqv Lists

Rakudo fix:
[Coke] shouldn't the subtest name be !eqv ? 17:58
IOninja Well, I'm testing `eqv`... the `!` is just a meta operator on it. I'd use `eqv` if we had cpm-nok; 18:01
Geth roast: a66cf6d98d | (Zoffix Znet)++ | S03-operators/eqv.t
Reword test to better indicate intent; [Coke]++
[Coke] ... I mean in the test title, not in the code. ah well. :)
IOninja I got you, but the test title was accurate and the code was a contortion to make `cmp-ok` succeed :)_ 18:04
[Coke] I disagree, but it's fine. 18:06
IOninja Depends on whether you read `Seq eqv List` as having implied `=== True` after it, I guess 18:08
[Coke] right. :) 18:14
(which i clearly do. :)
Geth rakudo/nom: 9d497e9dbc | (Elizabeth Mattijsen)++ | src/core/Rakudo/
Streamline R:In.MERGESORT* a bit

  - remove one level of nqp::stmts
  - make sure we always get a List with an IterationBuffer as reified
   (well, if we were given an nqp::list, it still might be an nqp::list)
  - remove some unneeded Mu's
jnthn Input welcome on $*OUT/$*ERR "inheritance": 22:46
TimToady: Wouldn't mind your take on ^^ also :) 22:47
In that area: I'm also considering whether to implement shell/run/Proc in terms of Proc::Async 22:48
They use the same libuv constructs under the hood, but we have to duplicately maintain a bunch of code in MoarVM 22:49
IOninja 'The docs are not arguably not wrong' wa 22:50
jnthn oops :)
IOninja :)
jnthn Got rid of the first not
I was gonna say "The docs are not wrong" which yeah, they ain't factually, but docs aren't just there to be factual :P
IOninja Sounds like a useful feature to me. 22:54
jnthn Time for me to get some rest :) 'night o/ 22:55
IOninja night