samcv MasterDuke, the docs page almost seems like we stole someone's table? 00:11
q, L, or ll interpret integer as C type "long long",
"unsigned long long", or "quad" (typically
64-bit integers)
or i guess this is for C compatibility? it seems odd though that we have this 00:12
geekosaur yes
samcv also the "these examples are subject to system-specific variation" on many of the examples
geekosaur, yes to which? 00:13
geekosaur various libc-s added their own extensions for 64 bit / long long before it got standardized, so variouys things support the most common add-ons
samcv ok
geekosaur and, more or less, any "printf" is going to use the ANSI C or libc one as its reference implementation 00:14
MasterDuke samcv: i think tbrowder copied it from the sprintf man page
geekosaur (some impls will stick to ANSI C, some will use glibc, etc.)
tbrowder samcv: no pride of authorship. i was trying to rationalize what works in p6 vs what is not available yet and what won't be available. i was going though actually trying each type but stopped further work at what you see. finishing is way down my TODO list at the moment. 00:18
samcv tbrowder++ 00:20
MasterDuke tbrowder: btw, i tried to run your file-reading benchmark yesterday, but it's still looking for some stuff in a Misc::Utils. anyway, i think Perl 6 should have gotten a bit faster since your last results 00:21
tbrowder: btw, adapted would have been a better word, i didn't mean to imply you hadn't put any thought into it 00:22
tbrowder yes, the reader is publicly rotten at the moment--i'm nearly ready to release some modules that replace the non-released Misc::Utils (which has been split into several more-focused modules based on critiques and suggestion from #perl6). but the last time i ran the reader (couple of weeks ago) there was no significant speedup vs. p5. 00:24
MasterDuke huh, must have been thinking of something else 00:26
TimToady . 01:10
yoleaux2 22 Jan 2017 23:52Z <MasterDuke> TimToady: do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
TimToady shrugs too
I can think of implementations where it might be costly to continually translate %0۳d to %03d 01:12
MasterDuke hm, two shrugs and one "that would be cool". those scales aren't moving to quickly 01:13
*too
TimToady it runs into the "trying to stuff a static language into a string" problem that p5 has with regexes (along with many other languages) 01:14
MasterDuke well, changing this line implements %0۳d github.com/perl6/nqp/blob/master/s...f.nqp#L161 01:15
`$st<size>.made` -> `nqp::radix(10, $st<size>.made, 0, 0)[0]`
TimToady well, that imposes a penalty on everyone, every time 01:16
if it could be attached to a failure test that doesn't sad-path the happy path, that would be better 01:17
MasterDuke something similar is done in a bunch of places to support Nd's
TimToady well, maybe it's that way internaly 01:18
*ally
MasterDuke you mean if it (some operation) fails, check for Nd and then restart with the 'value' instead of the Nd? 01:19
TimToady in general, I'd hope we attempt the fastest possible ASCII conversions, and if that fails, try again with Nd 01:20
since the fast majority of numbers will be written in ASCII
*vast
yes, restart
MasterDuke i don't think i've ever seen that idiom in the nqp/rakudo source 01:21
TimToady well, it's just an optimization, really, and optimization is cheating, which is technique :)
but guessing ASCII is gonna be a real good bet
MasterDuke i'll look around and see if that's done anywhere (and/or how to add it to where we go to nqp::radix right away) 01:23
TimToady and in general, I/O with base-10 numbers is worth putting a great deal of optimization effort towards
I understand python has a very fast hand-coded base-10 output routine 01:24
might be worth stealing
brokenchicken Does anyone know how come we return `Mu` here when key doesn't match? github.com/rakudo/rakudo/blob/e195...air.pm#L79 02:51
Since Mu isn't accepted by most operators, that results in LTA behaviour (as samcv just found out in #perl6)
m: my %h; dd %h<f> # Hashes, on the other hand return an Any 02:52
camelia rakudo-moar e195e5: OUTPUT«Any %h = Any␤»
TimToady it should probably return Nil, not Mu 02:56
brokenchicken does so
Geth kudo/nom: 9728c4a455 | (Zoffix Znet)++ | src/core/Pair.pm
Change value of Pair.AT-KEY on non-existent key

Currently the code returns a Mu, which isn't accepted by most operators and causes show-stopping and confusing errors.
Fix by returning a much more friendly Nil instead.
Discussion: irclog.perlgeek.de/perl6-dev/2017-...i_13971026
03:21
ast: 81a7ce94b3 | (Zoffix Znet)++ | S02-types/pair.t
Test Pair.AT-KEY for nonexistent key

Rakudo fix: github.com/rakudo/rakudo/commit/9728c4a455
p: b069919fb8 | (Samantha McVey)++ | src/HLL/Actions.nqp
Put single quotes around the character name on the error message

To make it more readable.
04:40
kudo/nom: 58226059b5 | (Samantha McVey)++ | src/Perl6/Actions.nqp
Use [] to delimit incorrect names in CURSOR.panic

  See discussion from here on back:
  irclog.perlgeek.de/perl6/2017-01-23#i_13971510
05:26
[Tux] This is Rakudo version 2017.01-25-g58226059b built on MoarVM version 2017.01-10-g76872dbe 07:38
csv-ip5xs 2.932
test 12.782
test-t 5.267
csv-parser 14.197
lizmat Files=1166, Tests=56204, 192 wallclock secs (11.46 usr 4.31 sys + 1160.80 cusr 110.93 csys = 1287.50 CPU) 10:29
yoleaux2 22 Jan 2017 23:52Z <MasterDuke> lizmat: do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
lizmat .tell MasterDuke seems like a logical continuation of the other Nd work you've done :-) 10:31
yoleaux2 lizmat: I'll pass your message to MasterDuke.
Geth kudo/nom: f230224d27 | (Elizabeth Mattijsen)++ | src/core/Backtrace.pm
Fix for RT #130630

Basically do not include any .nqp files in backtraces.
11:11
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130630
lizmat m: dd (|(0,1)).List.WHAT # sorta expected to see List there 11:27
camelia rakudo-moar 582260: OUTPUT«Slip␤»
timotimo m: say Slip ~~ List 11:28
camelia rakudo-moar 582260: OUTPUT«True␤»
timotimo m: dd (|(0, 1)).list.WHAT 11:29
camelia rakudo-moar 582260: OUTPUT«Slip␤»
lizmat timotimo: that last one I think is ok 11:33
but the .List one is not
timotimo OK! 11:34
Geth kudo/nom: 4d4822a76c | (Elizabeth Mattijsen)++ | src/core/Slip.pm
Make sure coercing a Slip to a List returns a List
11:41
kudo/nom: 8733aa5157 | (Elizabeth Mattijsen)++ | src/core/Rakudo/Internals.pm
Make sure trying to Empty a shaped array will die

my @a[2] = |(); # works my @b[2] = Empty; # "Cannot Empty a shaped array as its size is fixed"
11:55
timotimo neat. 11:56
Geth kudo/nom: 0ee3b7dabb | (Elizabeth Mattijsen)++ | src/core/Map.pm
Make Map.sort about 1.5x faster

Which also makes Map.perl/gist/Str quite a bit faster
12:41
kudo/nom: b7c6e73916 | (Elizabeth Mattijsen)++ | src/core/Any-iterable-methods.pm
Set up multi sub sort a bit smarter

This should prevent a lot of slipping for the most generic case.
13:01
|Tux| This is Rakudo version 2017.01-30-gb7c6e7391 built on MoarVM version 2017.01-10-g76872dbe 14:35
csv-ip5xs 2.888
test 12.429
test-t 4.990
csv-parser 13.896
I think my server was busy this morning
lizmat hmmm.. looks like we lost Geth again ? 16:08
Geth kudo/nom: c13d0a4c06 | (Elizabeth Mattijsen)++ | 2 files
Simplify Map/Hash.invert

No special candidate for object hashes is needed
lizmat ah, no, just slow :-) brokenchicken?? 16:09
brokenchicken lizmat: I noticed sometimes GitHub doesn't immediatelly send the web hook
lizmat: yeah, in the bot's terminal I see the hook arriving AFTER you said "we lost" :) 16:10
lizmat ah, ok, so it's not Geth about to die :-) 16:15
diakopter orrrr brokenchicken had to look at the terminal for the message to arrive 16:27
brokenchicken nope, it arrived before I looked :) 16:28
timotimo status.github.com/ - many of their timings have flared up a bit very recently, maybe that had something to do with it 16:42
Geth kudo/nom: 2f69be9144 | (Elizabeth Mattijsen)++ | 2 files
Simplify Map/Hash/TypedHash .pairs / .iterator mess

The base of iterating over an Associative class, should always be the .iterator method. All other things, such as .pairs, follow from that.
16:49
ast: 2dc968ec59 | (Elizabeth Mattijsen)++ | S09-hashes/objecthash.t
Fix too strict tests for empty hash listification
16:59
ast/6.c-errata: dfe905a8ce | (Elizabeth Mattijsen)++ | S09-hashes/objecthash.t
Fix too strict tests for empty hash listification
17:00
brokenchicken m: dd % .list 17:13
camelia rakudo-moar 2f69be: OUTPUT«().Seq␤»
brokenchicken m: dd % .List
camelia rakudo-moar 2f69be: OUTPUT«().Seq␤»
brokenchicken Should the title-case List return a List?
moritz yes 17:19
lizmat brokenchicken moritz : looking at that, please provide tests :-) 17:21
moritz m: use Test; plan 1; isa-ok {}.List, List; 17:22
camelia rakudo-moar 2f69be: OUTPUT«1..1␤not ok 1 - The object is-a 'List'␤␤# Failed test 'The object is-a 'List''␤# at <tmp> line 1␤# Actual type: Seq␤# Looks like you failed 1 test of 1␤»
moritz lizmat: ^^ test right there for you
brokenchicken huh 19:32
something crashed it :/ 19:33
Geth kudo/nom: 6dd542f6de | (Elizabeth Mattijsen)++ | 2 files
Make sure that Map/Hash.List returns a List
19:45
brokenchicken \o/ 19:56
wow much bigger fix than I'd expect it to be 19:57
travis-ci Rakudo build errored. Elizabeth Mattijsen 'Make sure that Map/Hash.List returns a List' 22:00
travis-ci.org/rakudo/rakudo/builds/194585906 github.com/rakudo/rakudo/compare/2...d542f6de3c
buggable [travis build above] ✓ All failures are due to timeout (2), missing build log (0), or GitHub connectivity (0).
samcv i'm trying to optimize string search speed for needles of length 2, but i'm running into an issue. some of the nqp tests fail. and i'm not sure how to diagnose, it just kind of hangs 22:18
well it seems to work fine and then it hangs, idk how to run gdb in a test harness
jnthn samcv: ps to find the ID of the hung process, gdb can be started with a process ID to attach to it 22:19
yoleaux2 22 Jan 2017 23:52Z <MasterDuke> jnthn: do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work?
samcv it freezes right after showing `ok 534 - literal match with quote`
when running prove -v -e 'nqp' ./t/qregex/01-qregex.t
thanks jnthn 22:20
jnthn Too tired to backlog here/#moarvm today; will try and catch up tomorrow :) 22:22
samcv hmm jnthn i'm not sure what is going on 22:33
gonna turn on debug and try 22:34
Geth ast: f22a9cc94b | (Zoffix Znet)++ | S12-class/attributes.t
Remove trailing whitespace
22:42
samcv yeah it's not failing on something in the code i wrote at least
Geth kudo/nom: 700a07747a | (Zoffix Znet)++ | src/Perl6/Actions.nqp
Fix .= initialization of attributes for types w/ `::` in name

The Perl6::World needs all namespace parts as separate strings in the Array, but currently it shoves the entire name as one item, which makes World fail to find the needed type.
Fix by splitting the name on `::` before handing it to World.
Bug find: irclog.perlgeek.de/perl6/2017-01-23#i_13976744
ast: afe080a028 | (Zoffix Znet)++ | S12-class/attributes.t
Test attributes can be initialized with .= when type name has `::`

Rakudo fix: github.com/rakudo/rakudo/commit/700a07747a Bug find: irclog.perlgeek.de/perl6/2017-01-23#i_13976744
lizmat and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/01/23/...4-welcome/ 22:51
good night, #perl6-dev! 23:02
jnthn 'night, lizmat 23:03
timotimo night lizmat :)
jnthn lizmat++ # another weekly 23:04
m: say (1..20).rotor(3) 23:05
camelia rakudo-moar 700a07: OUTPUT«((1 2 3) (4 5 6) (7 8 9) (10 11 12) (13 14 15) (16 17 18))␤»
jnthn m: say (1..20).batch(3)
camelia rakudo-moar 700a07: OUTPUT«((1 2 3) (4 5 6) (7 8 9) (10 11 12) (13 14 15) (16 17 18) (19 20))␤»
jnthn Why .batch?
m: say (1..20).rotor(3, :partial)
camelia rakudo-moar 700a07: OUTPUT«((1 2 3) (4 5 6) (7 8 9) (10 11 12) (13 14 15) (16 17 18) (19 20))␤»
jnthn rotor already does this, and plenty more besides 23:06
sleep time & 23:09
samcv who do I talk to about p6lanet syndication? 23:27
want to write an article but would like to set that up first
brokenchicken samcv: irclog.perlgeek.de/moarvm/2017-01-10#i_13891857 23:30
irclog.perlgeek.de/moarvm/2017-01-10#i_13891857 23:31
gdd
Well, the link is a few lines below there. I don't know why my stupid ass browser refuses to copy it. 23:32
samcv .ask stmuk can you syndicate my perl 6 blog o/ pl6net? feed cry.nu/feed.xml
yoleaux2 samcv: I'll pass your message to stmuk.
brokenchicken github.com/stmuk/pl6anet.org/blob/...perlanetrc
samcv oh 23:34
k adding
timotimo m: Supply.from-list(^20).rotor(3).tap(*.perl.say) 23:56
camelia rakudo-moar 700a07: OUTPUT«No such method 'tap' for invocant of type 'Seq'␤ in block <unit> at <tmp> line 1␤␤»
timotimo m: Supply.from-list(^20).batch(3).tap(*.perl.say) 23:58
camelia rakudo-moar 700a07: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in block <unit> at <tmp> line 1␤␤»
timotimo not sure why this is?
bisectable6: Supply.from-list(^20).batch(3).tap(*.perl.say) 23:59
bisectable6 timotimo, On both starting points (old=2015.12 new=700a077) the exit code is 1 and the output is identical as well
timotimo, Output on both points: Too many positionals passed; expected 1 argument but got 2␤ in block <unit> at /tmp/ndEkad2kMt line 1␤
timotimo oooh
it wants :elems(3)
m: Supply.from-list(^20).batch(:3elems).tap(*.perl.say)
camelia rakudo-moar 700a07: OUTPUT«[0, 1, 2]␤[3, 4, 5]␤[6, 7, 8]␤[9, 10, 11]␤[12, 13, 14]␤[15, 16, 17]␤[18, 19]␤»