travis-ci Rakudo build passed. Zoffix Znet 'Include default %SUB-MAIN-OPTS arguments in it; AlexDaniel++' 00:17
AlexDaniel huh
HelpBot[perl6] 00:22
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
HelpBot[perl6] 00:23
buggable [travis build above] ✓ All failures are due to timeout (0), missing build log (0), GitHub connectivity (3), or failed make test (0).
timotimo that wasn't a github connectivity issuethough? 00:38
it was the ppa:ubuntu-toolchain-r thing that blew up somehow? or whatever contains that repo thingie?
timotimo bedtime 00:39
AlexDaniel .tell dogbert2 I've updated everything. 00:45
yoleaux AlexDaniel: I'll pass your message to dogbert2.
AlexDaniel LHF right here: 00:50
HelpBot[perl6] I didn't feel like adding yet another category. It's up there in the "travis glitches" category 01:04
[Tux] This is Rakudo version 2017.06-113-ge2517f802 built on MoarVM version 2017.06-30-g389e9732 06:32
csv-ip5xs 2.716
test 12.541
test-t 4.196 - 4.326
csv-parser 12.806
lizmat Files=1208, Tests=63472, 214 wallclock secs (13.02 usr 4.71 sys + 1308.09 cusr 130.26 csys = 1456.08 CPU) 08:12
Geth nqp/master: 10 commits pushed by pmurias++
travis-ci NQP build failed. pmurias '[js] Fix beautification and running of node 09:15
pmurias is disabling test for the JVM backend 09:23
no super elegant way to do that, I guess I'll just implement those ops on the JVM backend 09:38
lizmat pmurias++ 09:39
Geth rakudo/nom: da6c6584f7 | (Elizabeth Mattijsen)++ | src/core/
Don't force MAIN-SUB-OPTS overhead on every startup

If a developer of a script wants to use MAIN-SUB-OPTS, she will have to create the lexicaly dynamic herself. This is in the nature of dynamic variables: there is no need for the core setting have them created automagically. This also sends a bad cognitive signal about dynamic variables.
This patch also for the time being adds temporary support for the old MAIN-ALLOW-NAMED-ANYWHERE, so that users of a newer rakudo with an older panda, will still work.
roast: 417f8e079b | (Elizabeth Mattijsen)++ | S06-other/main.t
Adapt SUB-MAIN-OPTS tests

The lexical dynamic should be created by the user.
lizmat hope I didn't upset Zoffix with this change 10:34
pmurias jnthn: should nqp::strcodes on the JVM treat NORMALIZE_NONE as NFC? the op docs say so but MoarVM just doesn't accept it 10:39
jnthn NONE would mean "don't do anything" 10:41
(Just get the original codepoints)
pmurias jnthn: we use NFG for strings on MoarVM and I assume we don't want to expose that right? 10:49
jnthn Expose what, exactly?
We never expose negative codepoints, no 10:50
At least, certainly not at Perl 6 level, but we don't do it really at VM level either
Geth roast: 408be80ff8 | (Zoffix Znet)++ (committed using GitHub Web editor) | S06-other/main.t
Remove no-longer needed test

The test tested the %*SUB-MAIN-OPTS key exists and it no longer does
pmurias nqp-m: nqp::strtocodes('str', nqp::null(), nqp::const::NORMALIZE_NONE) 10:54
camelia Invalid normalization form 0
at <tmp>:1 (<ephemeral file>:<mainline>)
from gen/moar/stage2/NQPHLL.nqp:1589 (/home/camelia/rakudo-m-inst-2/share/nqp/lib/NQPHLL.moarvm:eval)
from gen/moar/stage2/NQPHLL.nqp:1806 (/home/camelia/rakudo-m-inst-2/share/nq…
HelpBot[perl6] No, not upset
jnthn pmurias: Ah, it doesn't make sense in that context to use NORMALIZE_NONE, it was intended for when we have a bytestocodes op 10:55
lizmat works on the doc 10:57
pmurias jnthn: do we want nqp::normalizecodes to accept NONE? 10:59
jnthn Um 11:00
I guess it could return identity :)
That's the one that takes a codepoint array and returns a codepoint array, yes?
I mean, it's a pointless thing to do, so we don't do it :) 11:01
(From Perl 6 level)
lizmat hmmm looks like we lost geth 11:05
lizmat just committed
eh, 11:06
travis-ci NQP build passed. pmurias 'Update docs so that NORMALIZE_NONE to nqp::strtocodes is not allowed' 11:11
lizmat afk for a bit& 11:13
HelpBot[perl6] goes off the grid until Tuesday 11:52
[Coke] ~~ 11:54
timotimo .o( afk for a bitwise and ) 12:01
pmurias does data derived from the UNIDATA (machine readable unicode info) need to be particulary licencsed? 12:15
timotimo pre-encoding the string to a buf and writing instead of printing that makes it 2x faster 12:22
and yeah, this is most probably down to buffering, as perl5 writes 8k at once, and p6 writes 100 bytes at a time 12:24
jnthn Pass :buffer to open :) 12:28
timotimo oh, we already have that?
timotimo tries
jnthn Yeah, not the default yet
timotimo that's only a bit faster 12:29
26% time in encode-chars, 22.7% in print, 20.8% in <unit>, 14.9% time in <anon> (which is likely the inner loop that calls "print"), 6.5% in write-internal, 4.88% in pull-one, 3.9% in sink 12:30
pre-encoding the str in there makes a lot of difference, obviously 12:32
haha, is John Whitington jnthn's evil twin from another timeline? :D
there's like 3/4 overhead from the for loop alone or something like that 12:38
interestingly, the profler shows a call to sink from write (IO/, but i don't see the typical sink bbs in the speshlog for that 12:47
i might have an old speshlog, actually 12:48
yeah, no. i don't see a sink there
Geth rakudo/nom: 2d07e8a70c | (Elizabeth Mattijsen)++ | src/core/

if we have an explicit False in SUB-MAIN-OPTS<named-anywhere>
rakudo/nom: 385092270a | (Elizabeth Mattijsen)++ | src/core/
Make Iterable (+) Iterable about 20% faster

  - based on 1000 elem @a with itself
  - improvement mostly based on using about 1/3 of memory before
  - as only one final hash is created without 2 additional temporary ones
roast: 51fe30b56a | (Elizabeth Mattijsen)++ | S03-operators/addition.t
Add some more (+) tests

  - more than one pair with same key in Iterable
  - test for object hash vs normal hash
raschipi lizmat++ 14:49
lizmat afk again& 15:08
cog_ sz=-p=-[ 17:14
Geth roast: 385a2c3e61 | (Elizabeth Mattijsen)++ | S03-operators/addition.t
Add tests for infix:<(+)>() and infix:<(+)>(foo)
rakudo/nom: 779e6bb033 | (Elizabeth Mattijsen)++ | src/core/
Remove unnecessary tests for lack of parameters

These are all handled by specific candidates.
roast: a2c4696753 | (Elizabeth Mattijsen)++ | S03-operators/difference.t
Add tests for infix:<(-)>() and infix:<(-)>(foo)
lizmat m: dd ^10 (&) ^Inf # this feels wrong :-) 21:38
camelia set()
Geth roast: 3ab6c0b353 | (Elizabeth Mattijsen)++ | S03-operators/intersection.t
Add tests for infix:<(&)>() and infix:<(&)>(foo)

Also add tests for handling lazy lists, which for the moment still fail :-(
roast: cfbefc8b36 | (Elizabeth Mattijsen)++ | S03-operators/intersection.t
Activate now passing tests
roast: 7f75680c83 | (Elizabeth Mattijsen)++ | 2 files
Remove now superfluous (+) tests

Now done in S03-operators/addition.t
rakudo/nom: cb06ebacaa | (Elizabeth Mattijsen)++ | src/core/
Temporary fix for (&)'s handling of lazy lists
rakudo/nom: 04746490c9 | (Elizabeth Mattijsen)++ | src/core/
Make sure the (elem) family handles lazy lists

Before it would hang if it couldn't find the elem, such as:
   "a" (elem) ^Inf
It now fails with X::Cannot::Lazy
lizmat and this concludes my hacking for today 23:00
good night!
timotimo nite liz
jnthn 'night, lizmat++