dalek ast: 5844a89 | samcv++ | S15-nfg/ (5 files):
S15-nfg update mass-roundtrip-* for Unicode 9.0

The script is now updated to work with Unicode 9.0 as well.		 01:00
ast: f8070c4 | samcv++ | S15-nfg/mass-equality (2 files):
test-gen.p6's mass-equality.t generation has been moved to a new script

This was working fine for Unicode 9.0, so it was not messed with due to it being much more complex and involved rearranging the codepoints.
notviki samcv: this might be up your alley. Is that even a bug? doesn't look like that closing quote is the matching quote for the opener: rt.perl.org/Ticket/Display.html?id=130406 01:04
geekosaur was under the impression matching pairs weren't part of the spec... because they can depend on the locale 01:07
samcv let me look 01:11
m: Q„1“;
camelia rakudo-moar 1374fc: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Couldn't find terminator ” (corresponding „ was at line 1)␤at <tmp>:1␤------> Q„1“;⏏<EOL>␤ expecting any of:␤ ”␤»
samcv well that seems fine right. unless
yeah i don't think that 01:12
notviki Reading the code, it actually goes out to NQP land and uses our brackets string to find the matching losing delimiter
samcv yep
yeah it's not in bidi paired brackets or bidi mirroring glyphs 01:13
so it is not a proper matched bracket (/quote) 01:14
notviki m: Q„1”;
camelia rakudo-moar 1374fc: OUTPUT«WARNINGS for <tmp>:␤Useless use of constant string "1" in sink context (line 1)␤»
samcv m: "„".uniprop.say
camelia rakudo-moar 1374fc: OUTPUT«Ps␤»
samcv m: "„".uniname.say 01:15
camelia rakudo-moar 1374fc: OUTPUT«DOUBLE LOW-9 QUOTATION MARK␤»
samcv there is no matching thing for this.
almost certain
notviki bets $1000 samcv doesn't use 80-char linewidth
samcv yeah notviki there is no matching character for that
u: double quotation mark
unicodable6 samcv, U+00AB LEFT-POINTING DOUBLE ANGLE QUOTATION MARK [Pi] («)
samcv, U+00BB RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK [Pf] (»)
samcv, U+201C LEFT DOUBLE QUOTATION MARK [Pi] (“)
samcv, gist.github.com/987342005a219bbaf5...b968f36cbc
samcv yep 01:16
i remember because of before when i checked every single Ps and Pi thingy
m: "„".uniprop('Bidi_Mirrored').say 01:17
camelia rakudo-moar 1374fc: OUTPUT«False␤»
notviki i.imgur.com/kA9qztb.png :)
samcv well. it was like 100000000000000000 calumns wide before!
notviki :)
samcv all on one line :P
i will fit to 80 columns though
soonish. idk sometime today 01:18
notviki, about that bug report. tell them that Q quoting is not the same as quotation marks 01:19
and so take two opposite/matching delimiters. in most cases they overlap 01:20
like “ ” but not for this
notviki m: dd Q“x”
camelia rakudo-moar 1374fc: OUTPUT«"x"␤»
samcv oh they messaged me saying they got the CLA 01:21
notviki \o/
samcv should i forward that email to somebody
notviki In my case they notified jnthn and he gave me the commit bit
b2gills m: say „1“;
camelia rakudo-moar 1374fc: OUTPUT«1␤»
samcv You can let the Rakudo admins know that we've received the CLA.
well i am letting you guys know :P unknown whether they sent anything
it was Allison which replied to my email 01:22
notviki passes samcv the torch
I'm no longer the newest core member \o/
samcv \o/
dalek kudo/nom: 1ee24cc | (Zoffix Znet)++ | src/core/List.pm:
Fix spurious warnings in List.reduce

When the mapper sub has associativity set, the %prec ends up having stuff (so the first conditional passes), but the `prec` key is still absent. So when we infix:<eq> on it, it triggers warnings about stringifying uninitialzed values.
Fix by checking if it's inited first.
Bugfind: irclog.perlgeek.de/perl6/2016-12-26#i_13799375		 02:06
ast: 261c2c3 | (Zoffix Znet)++ | packages/Test/Util.pm:
Add doesn't-warn() utility test sub

Checks whether a warning is emitted. This lets us test for warnings faster than by using is_run to fire up an entire perl6 for us.		 02:11
ast: 27ddcc8 | (Zoffix Znet)++ | S32-list/reduce.t:
Remove trailing whitespace
ast: b2e46ab | (Zoffix Znet)++ | S32-list/reduce.t:
Test .reduce doesn't produce spurious warnings

9418115 | samcv++ | S15-unicode-information/uniprop.t: Fudge some more tests for uniprop
notviki stares down dalek
hm, I can teach huggable to take over for him when he fails like that 02:18
t/spec/S15-unicode-information/uniprop.rakudo.moar (Wstat: 768 Tests: 152 Failed: 3) 02:23
Failed tests: 121-123
Non-zero exit status: 3
samcv do i need to fudge some more? 02:24
line numbers?
actually nvm i will do it 02:25
samcv needs to get multiple installations of rakudo installed
notviki Rakudobrew is just for that: github.com/tadzik/rakudobrew 02:26
samcv notviki, how do you switch between them easiest notviki 02:30
ok. passes now :)
notviki By "multiple" you mean different commits? I think it's rakudobrew switch 02:36
notviki doesn't actually use multiple versions
samcv ack ok my changes to moarvm mess up one of the tests 02:41
oddly somehow it's this rt.perl.org/Ticket/Display.html?id=125190 03:04
hm 03:05
ack so it wosn't any of my changes at all. 03:16
hmm well i got it fixed. so that is all good 03:21
er maybe not. this is really weird. will need to investigate. this is quite odd 03:24
m: say ' ' ~~ /<:space>/ 03:56
camelia rakudo-moar 1ee24c: OUTPUT«｢ ｣␤»
samcv m: say ' ' ~~ /<:Space>/
camelia rakudo-moar 1ee24c: OUTPUT«｢ ｣␤»
samcv wtf. so for some reason it's failing for :space but not :Space, which works fine 03:57
ugh what gives. uniprop-bool for 'space' returns true, so it seems to be fine. but it fails that test 04:01
with the regex
dalek kudo/nom: b01dfcd | (Zoffix Znet)++ | src/Perl6/Actions.nqp:
Warn when using Bool:D as type constraint in signatures

Fixes RT#130182: rt.perl.org/Ticket/Display.html?id=130182		 06:34
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130182
dalek ast: 7095892 | (Zoffix Znet)++ | S06-signature/types.t:
Test Bool:D as literals in signatures warn

RT#130182: rt.perl.org/Ticket/Display.html?id=130182 Rakudo fix: github.com/rakudo/rakudo/commit/b01dfcda90		 06:35
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130182
gfldex .u „ 09:57
yoleaux2 U+201E DOUBLE LOW-9 QUOTATION MARK [Ps] („)
gfldex .u “ 09:58
yoleaux2 U+201C LEFT DOUBLE QUOTATION MARK [Pi] (“)
gfldex .u “
yoleaux2 U+201C LEFT DOUBLE QUOTATION MARK [Pi] (“)
gfldex notviki: this will come up again because of de.wikipedia.org/wiki/Anf%C3%BChru...e_Sprachen 09:59
loads of eastern european languages use them 10:00
samcv the main question becomes, do we support unicode matching brackets + mirroring glyphs that are open/closers AND all the quoting constructs we allow? 10:03
at the moment we only allow matching brackets/mirroring glyphs and also characters that are the same open and same close
gfldex m: say „we do, but not for Q“. 10:04
camelia rakudo-moar b01dfc: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Malformed postfix call␤at <tmp>:1␤------> say „we do, but not for Q“.⏏<EOL>␤»
samcv if we add that one, we should add every one of the quotation marks
m: say ”test”; say Q”test”
gfldex m: say „we do, but not for Q.“
camelia rakudo-moar b01dfc: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Use of a closing delimiter for an opener is reserved␤at <tmp>:1␤------> say ”test”; say Q⏏”test”␤»
rakudo-moar b01dfc: OUTPUT«we do, but not for Q.␤»
samcv yeah exactly
gfldex i'm cool with that, just need to doc it.
samcv i personally favor just going with brackets/mirroring glyphs, but being consistent is the most important 10:05
it's about time for my bed. night everybody :) 10:12
dalek kudo/nom: 2496963 | lizmat++ | src/core/Str.pm:
Re-imagine Str.split(<a b c>) some more

  - 10% (small string, 1 hit each) to 40% (larger, 100 hits each) faster
  - don't use nqp::p6sort anymore, instead use R:I.MERGESORT-REIFIED-LIST-WITH
  - so we don't need to setup indices + we have mapping simplified
  - don't copy values when limiting, just limit the original result		 14:11
lizmat this should probably also be noticeable in test-t ^^^
only one more step towards the elimination of nqp::p6sort
but that's for later today 14:12
afk&
notviki m: dd 16:37
camelia rakudo-moar 249696: OUTPUT«block <unit>␤»
notviki m: dd ‘…‚
camelia rakudo-moar 249696: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Unable to parse expression in curly single quotes; couldn't find final "’" ␤at <tmp>:1␤------> dd ‘…‚⏏<EOL>␤ expecting any of:␤ argument list␤ curly single quotes…»
notviki m: dd “…„ 16:38
camelia rakudo-moar 249696: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Unable to parse expression in curly double quotes; couldn't find final '”' ␤at <tmp>:1␤------> dd “…„⏏<EOL>␤ expecting any of:␤ argument list␤ curly double quotes…»
notviki If there's anything we can learn from supporting U+2212 is to say "no" once in a while. 16:39
[Tux] Nice ♥ ! 16:58
This is Rakudo version 2016.12-102-g2496963ea built on MoarVM version 2016.12-19-ge2ffc358
csv-ip5xs 3.057
test 14.270
test-t 5.328
csv-parser 15.208
notviki \o/ 17:05
buggable: speed
buggable notviki, ▅▅▅▆▅▅▅▄▅▅▆▅▅▅▆▄▄▄▄▄▅▄▄▅▄█▃▃▄▄▃▄▄▃▄▃▄▄▃▄▄▄▂▃▂▂▂▂▁▁ data for 2016-12-04–2016-12-26; variance: 5.328s–7.592s
notviki :D
hmm... I got $node.node.CURSOR.worry("Useless use of $sym in sink context"); call in unwanted() and I added debug prints to see that I'm hitting that conditional, but no warning gets printed 18:02
mst notviki: your debug print is going to stderr or an autoflushing filehandle right? 18:08
(probably a stupid question but I've seen/done it so many times)
notviki Yeah to stderr. Why? 18:10
m: 1...2
camelia rakudo-moar 249696: OUTPUT«Potential difficulties:␤ Useless use of ... in sink context␤ at <tmp>:1␤ ------> 1...⏏2␤»
notviki It's the same conditional that prints ^ that warning
I'm trying to make
m: 1 but 2
camelia ( no output )
notviki Warn and it doesn't... so I'm missing something else that needs to be done 18:11
lizmat .tell samcv did you forget to push Missing test file: t/spec/S15-nfg/mass-chars.t ? 18:53
yoleaux2 lizmat: I'll pass your message to samcv.
dalek kudo/nom: 7ac0455 | lizmat++ | src/core/Rakudo/Internals.pm:
Introducing R:I.MERGESORT-REIFIED-LIST-AS

  - sort a reified list using a Schwartzian transform		 19:05
kudo/nom: def5262 | lizmat++ | src/core/Any-iterable-methods.pm:
Any.sort(&by) uses R:I.MERGESORT-REIFIED-LIST-AS

  - no need to set up indexes (MERGESORT-REIFIED-LIST-AS does that)
  - otherwise same algorithm as MERGESORT-REIFIED-LIST
  - about 40% faster for a 100-element list
kudo/nom: 3a1fb3a | lizmat++ | src/vm/moar/Perl6/Ops.nqp:
RIP nqp::p6sort

Thanks for all the hard work!		 19:06
notviki lizmat: it was purposefully deleted: github.com/perl6/roast/commit/3baa...d1f30dda15 19:40
And I .tell'ed jnthn to review and OK that change. 19:41
And if it is, we'll just stick a dummy file in place, I guess.... to make make spectest quiet about missing files :\ 19:42
(we can't just remove it from t/spectest.data, because it's in 6.c-errata) 19:43
samcv uhm is this a bug rt.perl.org/Ticket/Display.html?id=117683 20:11
yoleaux2 18:20Z <jnthn> samcv: About test-gen.p6 it was iirc written before some fairly major changes (including the Great List Refactor), and it's entirely possible the generator script is busted as a result. It was run against UCD 8.0 originally, so certainly produced passing tests against that. So I suggest "use UCD 8.0 and fix it to prdouce the same output first" would be one way to go.
18:21Z <jnthn> samcv: Will try and catch up with your MoarVM PR and other Unicode bits that may want my input soon too; getting some much-needed rest these last days. :-)
18:53Z <lizmat> samcv: did you forget to push Missing test file: t/spec/S15-nfg/mass-chars.t ?
samcv lizmat, i removed it
.tell lizmat it has been removed see my explanation here: github.com/perl6/roast/commit/3baa...d1f30dda15 20:12
yoleaux2 samcv: I'll pass your message to lizmat.
samcv and all the fudged tests we were failing from masschars.t, the reason was just the script generating them making bad assumptions. so using the GraphemeBreakTest, Unicode actually provides HOW graphemes are broken up. instead of us trying to do it ourselves, and so the fudged tests in GraphemeBreakTest are actually correct (aka SHOULD be being passed but don't) instead of tests which shouldn't pass, but the script generated them and they were bad 20:16
notviki, i can always write something new for masschars.t if you want, if it becomes needed, though. or maybe write something to test how we further break up the characters (instead of just checking .chars). going more deeply into the GraphemeBreakTest and checking each grapheme boundary instead of our total grapheme count 20:21
dogbert2 samcv: shouldn't t/spec/S15-nfg/mass-chars.t be removed from 'spectest.data' as well? 20:22
samcv no, becuase it's still in roast
for 6.c-errata 20:23
dogbert2 so the 'Missing test file: t/spec/S15-nfg/mass-chars.t' when running 'make spectest' is ok then?
samcv yes :)
read my commit here github.com/perl6/roast/commit/3baa...d1f30dda15 20:24
and because the graphemebreaktest actually tells US how they break, instead of us creating the number of characters it 'should be' using our own rules on the fly on test creation 20:25
which is not a very forward compatible way to do it
well and kind of bad regardless. since we need to take a lot more into account than one unicode property, plus the rules could change in the future 20:26
even if we did check all the rules _now_ on test creation, they could change and then we'd have to rewrite the generator to create the write 'answers' much better to rely on unicode's answers instead
dalek ast: de2fe04 | (Zoffix Znet)++ | S15-nfg/mass-chars.t:
Silence unwated missing file warnings
 20:28
samcv notviki++
dalek kudo/nom: 1e54371 | lizmat++ | src/core/List.pm:
Add List.sort(&by) candidate

On reified lists, it should not have to do a copy of the list being sorted. Gives about a 5% gain on reified List/Array.
dogbert2 Zoff, eh I meant notviki++ to the rescue :) 20:31
dalek kudo/nom: a7bf1ea | lizmat++ | src/core/List.pm:
Oops, R:I.MRLA is the one that doesn't need cloning

R:I.MRLW does need a clone to work with, although no spectest failed without the clone, oddly enough.		 20:37
samcv somehow I missed seeing this ticket. thanks notviki github.com/perl6/atom-language-perl6/issues/29 20:41
fixing now
i've got a good workflow going for the highlighter, wrote a unescape.p6 that takes stdin and will unescape and print out the cson. copy the text: xclip -o | unescape.p6 | xclip 20:43
and then it's on my clipboard unescaped :)
then have one to re-escape it once i'm done getting it working
notviki notices TimToady got a fancy hostmask for Christmas 20:48
notviki leaves to nurse a burn out 22:05
lizmat wonders whether she should mention www.reddit.com/r/perl/comments/5jk...h=b90d1568 in the P6W 22:19
yoleaux2 20:12Z <samcv> lizmat: it has been removed see my explanation here: github.com/perl6/roast/commit/3baa...d1f30dda15
samcv wow that's harsh
that reddit
lizmat decided against it 22:21
timotimo there was some extra text to it or something? 22:26
lizmat timotimo: yes, there was, but that was removed by the admins 22:43
but the title is still visible, and says enough, really
and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2016/12/26/...s-the-end/ 22:48
notviki lizmat++ good weekly. 22:56
lizmat: FWIW MasterDuke's 10% improvement is *on top* of the 40% improvement to perl6 -ne '' he did: github.com/rakudo/rakudo/pull/960
lizmat notviki: added, thanks! 22:58
notviki lizmat++ 22:59
samcv arghhhhhhh
0xF.uniprop-bool('space') #> true; 0xF.uniprop-bool('White_Space') #> False 23:00
goddamit space property. always flipping which is broken

Powered by Cro and the Raku® Programming Language.

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

Thank you!