»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:01 FROGGS left
Mouq github.com/sillymoose/Perl6-One-Li...t-64404187 00:08
raydiak timotimo: you could add a layer of indirection between your module's values and sdl's values, then rearrange the values attached to your public enums so all conflicting names have the same value :P
Mouq Is there a simpler way to write `$_ ~= "\n"` that I don't know about? 00:09
Having `~="\n"` seems like it would violate TTIAR
timotimo m: sub postfix:<~="\n">($a is rw) { $a = $a ~ "\n" }; my $a = "hooray!"; $a~="\n"; print $a.perl 00:12
camelia rakudo-moar e1aa07: OUTPUT«"hooray!\n"»
00:14 lizmat joined
Mouq m: sub prefix:<~=> ($s) { CALLER::<$_> ~= $s }; $_ = "hooray"; ~="\n"; .print # more like.. :) 00:14
camelia rakudo-moar e1aa07: OUTPUT«hooray␤»
00:14 exixt is now known as EXIXT
TimToady m: sub postfix:<n> ($x) { $x ~ "\n" }; print 42\n; 00:17
camelia rakudo-moar e1aa07: OUTPUT«42␤»
TimToady just don't trying append a tab that way :)
s/ing// 00:18
er, s/ing/ to/
Mouq m: sub postfix:<n> ($x) { $x ~ "\n" }; print 42\i\n; 00:19
camelia rakudo-moar e1aa07: OUTPUT«0+42i␤»
00:30 ssqq joined 00:33 perturbation left 00:39 denis_boyun_ left 00:44 jakesyl left 00:56 KCL_ joined
dalek p: ba888b5 | TimToady++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
one test for special cases; if -> switch; fix tabs

Seems to run about 1.5% faster.		 01:00
01:01 bjz joined 01:04 anaeem1 joined 01:05 anaeem1 left
Mouq m: my $g = (^4 Z=> (1..16).pick(4)).BagHash;say $g.fmt("%d: %d"); say "Total ", [+] $g.values; say $g.pick(*).uniq for ^30; 01:05
camelia rakudo-moar e1aa07: OUTPUT«0: 7␤1: 14␤2: 8␤3: 1␤Total 30␤0 1 2 3␤3 0 1 2␤0 1 2 3␤0 2 1 3␤1 2 0 3␤0 1 3 2␤0 1 2 3␤0 1 2 3␤1 2 0 3␤0 2 1 3␤0 1 2 3␤1 0 2 3␤2 1 0 3␤0 2 1 3␤1 3 2 0␤2 1 0 3␤1 2 0 3␤2 0 1 3␤2 1 0 3␤1 2 0 3␤1 0 2 3␤0 1 2 …»
01:05 anaeem1 joined
Mouq ^^ One-line version of github.com/perl6/perl6-examples/bl...-731696.pl 01:05
TimToady good thing my window is so wide... 01:08
01:10 anaeem1 left
TimToady that 1.5% faster was the timing on the parse phase of the setting, btw 01:12
timotimo cute
more of that? :)
TimToady overall build time was actually about 5% faster 01:13
but not very scientifically done
timotimo so switch/case is faster than if/else in java? 01:15
arnsholt Probably 01:18
switch compiles to a jump table, so if there are many branches, I wouldn't be surprised
dalek p: 489bd97 | TimToady++ | src/vm/parrot/ops/nqp.ops:
one test for special cases; if -> switch; fix tabs

  (same for parrot as we just did for jvm)		 01:20
01:22 colomon joined
arnsholt Oh yeah, that switch there is probably going to help I think. At least if it's kinda hot (like it looks like it is) 01:22
TimToady we still need to optimize switches in P6 that way
yes, the NFA is definitely hot code
arnsholt Yeah. That's trickier though, since P6 allows more complicated things in the cases than C and Java 01:23
TimToady however, we've been very careful to define smartmatching to depend only on the type of the "when"
so 'when 1' is always integer comparison 01:24
hence optimizable
arnsholt True, true
TimToady also strings, where you could dispatch based on, say, the first character
01:25 kst` is now known as kst
TimToady of course, if there are multiple entries with the same first letter, it would devolve to the current situation of testing each of those 01:25
so it would be good to group them
one could go as far as to generate a trie of all the strings, I suppose 01:26
01:26 dayangkun joined
TimToady but generally by the time you've gone to that much trouble, you should have been using polymorphism on method names instead 01:26
arnsholt Yeah, or something more clever in general
timotimo will foo[1]['bar'] += 10 create foo[1]['bar'] if foo[1] is a dictionary that does not include 'bar' yet?
most probably not :\ 01:27
TimToady well, I don't think either python or ruby do autoviv the way perl does
that's kinda one of perl's distinctives
arnsholt Standard Python dicts won't, no 01:28
There's defaultdict though, if you want that behaviour
TimToady even if perl5 botches it in spots
01:32 vike left 01:37 Akagi201 left 01:41 bjz left
Mouq github.com/perl6/perl6-examples/bl...ut/RESULTS 01:48
timotimo 2009 01:50
i hope we've improved a bit since then
Mouq A lot :) 01:51
And this is without compact arrays, and is compared to p5's greased-up regexes 01:57
timotimo did you calculate some up to date values
Mouq N-Body: Was: 550x slower than p5. Is: 47.3x. Is, with s/Num //: 27.8x 02:02
Although it looks like p5 takes more of a lead as N increases 02:03
Mouq finds it LTA that slurp() no longer works... :/ 02:04
colomon slurp in p5 no longer works? 02:05
Mouq Regex-DNA: Was 410x, is 16.9x
in p6
m: say slurp
camelia rakudo-moar e1aa07: OUTPUT«set encoding requires an object with REPR MVMOSHandle␤ in method encoding at src/gen/m-CORE.setting:15900␤ in method slurp-rest at src/gen/m-CORE.setting:15855␤ in sub slurp at src/gen/m-CORE.setting:16606␤ in block <unit> at /tmp/plB0OXczaV:1…» 02:06
Mouq j: say slurp
camelia rakudo-jvm e1aa07: OUTPUT«Can't open perl script "/home/camelia/jvm-rakudo/eval-client.pl": No such file or directory␤»
Mouq p: say slurp
camelia rakudo-parrot e1aa07: OUTPUT«No such method 'encoding' for invocant of type 'Any'␤ in method encoding at gen/parrot/CORE.setting:15822␤ in method slurp-rest at gen/parrot/CORE.setting:15774␤ in sub slurp at gen/parrot/CORE.setting:16514␤ in sub slurp at gen/parrot/CORE.s…»
Mouq K-Nucleotide: Was 200x, is 23.7x 02:07
colomon "/home/colomon/.ssh/config".IO.slurp works fine for me, though my Rakudo is a day or two old
Mouq Wait 02:08
Yeah
colomon: Yeah, but just plain 'ol `slurp` to get STDIN seems borked
Revcomp: Was 1600x, is 38.4x 02:10
colomon yeah, you're right. wonder if the rakudobug has been reported? it's clearly a bug...
Mouq Fasta: Was 590x, is 15x 02:11
Oh, wait 02:12
Well, I just realized I'm on a Rakudo from October 02:13
But if camelia fails it, it's definitely still a bug :P
japhb Mouq: Are these with no types, native types (num), or high-level types (Num)? 02:18
Mouq The timings? No native types. High-level types are used in repo's current N-Body impl and slow it down 02:20
TimToady well, Num is really not a very high-level type, just a num in disguise 02:25
Int is rather further from int, I think 02:26
dalek p: ce191a8 | (Timo Paulssen)++ | / (2 files):
show allocations split into interpd/speshd/jitted frames
 02:34
timotimo ^- enjoy
raydiak timotimo++ 02:35
02:38 dj_goku joined 02:39 adu joined, cognominal left 02:40 jimmy_ joined
colomon Mouq: yeah, the set encoding requires an object with REPR MVMOSHandle was the error I was getting 02:43
02:45 ilbot3 left 02:46 perturbation joined, adu left, ilbot3 joined 02:47 adu joined, cognominal joined 02:48 KCL joined 02:50 chenryn joined 02:51 KCL_ left, Mouq left 02:54 jimmy_ is now known as JimmyZ, dayangkun left
timotimo i hope it's correct 03:00
dj_goku does anyone know a good way to contact FROGGS? 03:02
timotimo if you .tell FROGGS blah blah, he'll get it the next time he writes on irc 03:03
colomon doesn't know a method better than invoking his name here….
timotimo otherwise [email@hidden.address]
dj_goku .tell FROGGS testers.perl6.org front-end seems to be down. :(
yoleaux dj_goku: I'll pass your message to FROGGS.
timotimo cpandatesters.org/ seems to work 03:04
dj_goku hmm maybe it is my internet?
because that didn't work either.
timotimo oh
dj_goku hmm from my vps it does.
weird they both work from my vps. 03:05
03:05 kurahaupo joined
dj_goku lol quit safar and now it works. 03:06
:(
s/safar/safari/ 03:07
03:20 noganex_ joined 03:22 noganex left 03:24 perturbation left 03:34 vike joined 03:44 Mso150 joined 03:48 adu left 03:51 Mso150 left
raydiak why don't we have something that autothreads like a junction and is introspectable like a set? never collapses, but coerces to a junction w/explicit .any/.all/.one/.none, otherwise maybe behaves like a list in bool context (.elems.Bool) 03:55
04:05 Mouq joined 04:09 raiph joined 04:13 rmgk left 04:14 rmgk joined 04:16 zakharyas joined 04:25 vike left 04:40 chenryn left 04:54 KCL_ joined 04:55 chenryn joined 04:56 KCL left 04:58 kurahaupo left
TimToady
.oO(and a pony)
 05:01
05:02 kaleem joined 05:04 anaeem1_ joined 05:08 anaeem1_ left
ssqq write *use v6* or ignore it, Have any difference? 05:10
05:12 mr-foobar left 05:13 zakharyas left 05:17 kaleem left 05:36 adu joined 05:49 kurahaupo joined 05:52 dayangkun joined 05:55 raiph left 05:58 kaare_ joined
moritz good morning 06:10
ssqq: with 'use v6;', you get a much better error message if you accidentally run it with perl 5
ssqq: that has happened to me quite a few times :-)
also, it makes it easier for tools (like syntax hilighters) to reliably determine the language 06:11
JimmyZ ssqq: 好久不见
ssqq JimmyZ: 好久不见。 06:29
moritz: I found CORE:.kyes have not change when ignore *use v6* 06:30
06:33 dayangkun left 06:49 anaeem1_ joined 06:50 adu left 06:57 kurahaupo left, ssqq left 06:58 telex left 07:00 kurahaupo joined, telex joined 07:01 kaleem joined 07:07 denis_boyun joined 07:10 salv0 left 07:25 blackbolt joined 07:26 salv0 joined 07:27 gfldex joined 07:36 jluis joined
dalek c: cbcd9b0 | Mouq++ | lib/Language/operators.pod:
Get rid of a TODO for dotty ops
 07:39
07:47 vti joined 07:49 [Sno] left 07:51 abraxxa left 08:00 FROGGS joined 08:05 gfldex left 08:09 Mouq left
masak mornin', #perl6 08:11
08:13 blackbolt left
lizmat morning masak 08:13
lizmat is about to commute to Lyon 08:14
for the Patch -p2 Hackathon
moritz
.oO( the Lyon sleeps tonight )
lizmat
.oO( I would by Lyon if I said that was wrong )
raydiak Lyon in green pastures? 08:16
Lyon battery in my laptop?
lizmat hopes it won't be freezing in Lyon
.oO( I don't like your Lyon Ice )
lizmat must stop with bad Lyon puns 08:17
08:17 Alina-malina left 08:18 virtualsue joined
masak .oO( I still have a milLyon of them left! ) 08:19
08:20 bjz joined, Alina-malina joined 08:22 rurban joined
raydiak g'night #perl6 08:25
08:27 kjs_ joined
masak 'night, antipodal raydiak 08:29
08:29 Mso150 joined 08:30 darutoko joined 08:42 Mouq joined 08:43 [Sno] joined 08:50 ptc_p6 joined 08:54 woolfy left 09:01 Alina-malina left 09:05 Alina-malina joined 09:07 Mouq left, rindolf joined 09:10 kjs_ left, virtualsue left
dalek kudo/newio: 124cab4 | (Elizabeth Mattijsen)++ | src/core/PIO.pm:
First version of PIO role (untested as yet)
 09:10
kudo/newio: 8af1338 | (Elizabeth Mattijsen)++ | src/core/IO/Pipe.pm:
IO::Pipe first consumer of PIO role
sven_123 perl6: say 3+4
camelia rakudo-{parrot,moar} e1aa07: OUTPUT«7␤»
lizmat commute to Lyon& 09:11
09:11 lizmat left 09:15 fhelmberger joined
FROGGS m: say 1, 2, 4 ... 32 # sven_123: it can do pretty nice things :o) 09:19
yoleaux 03:03Z <dj_goku> FROGGS: testers.perl6.org front-end seems to be down. :(
camelia rakudo-moar e1aa07: OUTPUT«1 2 4 8 16 32␤»
FROGGS damn
JimmyZ m: say 1, 2, 3 .. 10 09:20
camelia rakudo-moar e1aa07: OUTPUT«123..10␤»
JimmyZ m: say 1, 2, 3 ... 10
camelia rakudo-moar e1aa07: OUTPUT«1 2 3 4 5 6 7 8 9 10␤»
JimmyZ m: say 1, 2, 5 ... 10
camelia rakudo-moar e1aa07: OUTPUT«===SORRY!===␤Cannot assign to a readonly variable or a value␤»
FROGGS eww
that sounds like it fails to generate the exception or so
09:21 sqirrel__ joined
FROGGS dj_goku: thanks, I restarted it 09:21
JimmyZ aye
09:21 bjz left
FROGGS p: say 1, 2, 5 ... 10 09:21
camelia rakudo-parrot e1aa07: OUTPUT«Cannot assign to a readonly variable or a value␤ in block at gen/parrot/CORE.setting:19586␤ in any coro at gen/parrot/CORE.setting:8486␤ in method reify at gen/parrot/CORE.setting:8467␤ in block at gen/parrot/CORE.setting:8177␤ in method…»
JimmyZ n: say 1, 2, 5 ... 10 09:22
camelia niecza v24-109-g48a8de3: OUTPUT«Unhandled exception: Unable to figure out pattern of series␤ at /home/camelia/niecza/lib/CORE.setting line 1536 (die @ 5) ␤ at /home/camelia/niecza/lib/CORE.setting line 3423 (get-next-closure @ 37) ␤ at /home/camelia/niecza/lib/CORE.setting …»
FROGGS m: say 1, 2 ... 32, 33 ... 40
camelia rakudo-moar e1aa07: OUTPUT«1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40␤»
FROGGS m: say 1, 2, 4 ... 32, 33 ... 40
camelia rakudo-moar e1aa07: OUTPUT«1 2 4 8 16 32 33 34 35 36 37 38 39 40␤»
FROGGS well, at least that still works...
09:22 bjz joined
FROGGS I'll rakudobug the other issue 09:22
JimmyZ Thanks :) 09:23
09:24 Alina-malina left 09:29 vike joined
sven_123 FROGGS - is there a big difference between 'perl6:' and 'm:'? 09:30
masak m: say 4 09:31
camelia rakudo-moar e1aa07: OUTPUT«4␤»
masak perl6: say 4
camelia rakudo-{parrot,moar} e1aa07: OUTPUT«4␤»
masak p: say 5
camelia rakudo-parrot e1aa07: OUTPUT«5␤»
masak sven_123: hope that answes it.
09:32 dakkar joined
JimmyZ time perl6 -e 'my int $n = 0; loop { $n = $n +1; if $n == 10000000 { exit; } }' # real 0m0.934s 09:34
time php -r '$n =0; for(;;) { if($n++ == 10000000) { exit; }}' # real 0m0.490s
sven_123 actually not really - i mean one is the parrot-thing and the other is moar-vm, i read a bit about that, but i don't know enough, to know the real differences; guess mostly it doesn't matter, I just asked, because froggs wrote me an example and used m:
JimmyZ time perl6 -e 'my int $n = 0; loop { 10000000.rand; $n = $n +1; if $n == 10000000 { exit; } }' # real 0m10.562s 09:35
time php -r '$n =0; for(;;) { rand(1, 100000000); if($n++ == 10000000) { exit; }}' #real 0m4.185s
moritz sven_123: 'm:' is faster, uses less memory, and supports concurrency 09:36
masak m: sub stick-figure { my ($head, $arm, $legs) = (577 * 3).base(16).comb; my $body = [~] $arm, $head, $arm, $legs; chr :16($body) }; say stick-figure
camelia rakudo-moar e1aa07: OUTPUT«웃␤»
masak :D 09:37
masak hopes he just created a new type of obfuscated Perl 6 code category
sven_123 thx :) 09:38
09:44 zakharyas joined 09:53 pecastro joined 09:55 ptc_p6 left 09:56 ptc_p6 joined 09:57 EXIXT is now known as exixt 10:00 Ugator joined 10:03 virtualsue joined
JimmyZ .tell timotimo perl6 --profile -e 'my int $n := 0; loop { $n := $n +1; if $n == 10000000 { exit; } }' has 10000001 Scalar Allocations, It will be nice to avoid it 10:09
yoleaux JimmyZ: I'll pass your message to timotimo.
JimmyZ .tell timotimo please igonre ':'
yoleaux JimmyZ: I'll pass your message to timotimo.
10:11 risou is now known as risou_awy 10:12 pecastro left, risou_awy is now known as risou 10:14 cognominal left
JimmyZ .tell timotimo and I see Exclusive Time on exit is 18446744073709.23ms .. 10:16
yoleaux JimmyZ: I'll pass your message to timotimo.
10:21 chenryn left
Ulti masak that reminds me of a code golf I did in Java codegolf.stackexchange.com/question...3401#23401 10:23
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...l?id=23401
10:25 kjs_ joined 10:28 exixt is now known as EXIXT 10:29 JimmyZ left 10:30 EXIXT is now known as exixt 10:34 zakharyas left 10:36 sunnavy left, sqirrel__ left 10:38 daxim joined 10:41 chenryn joined 10:44 pecastro joined 10:49 Mso150 left 10:52 Mouq joined 10:56 Mouq left 11:00 anaeem___ joined 11:03 anaeem1_ left 11:13 sunnavy joined 11:25 pecastro left 11:27 pecastro joined
masak Ulti: writing that stick figure was fun. I'm pondering writing a (short) blog post about it. :) 11:28
masak ponders what it'd take to eliminate the synopsebot false positive above 11:35
either "is this #NNNNN thing not part of a URL?" or "was there whitespace or BOL before the '#'?" 11:37
11:45 pecastro left 11:49 bloonix left 11:54 chenryn left
tadzik well, we can have a test suite :) 11:55
12:06 pmurias joined
pmurias arnsholt: it seems that it's not that variables are only declared when set but that python keeps track if they were initialized before use or not 12:08
12:12 pecastro joined 12:35 sqirrel__ joined
arnsholt pmurias: Yeah, the relevant piece of spec is docs.python.org/3/reference/execut...nd-binding 12:40
It seems to me that this comes from the easy way to implement this in a simple interpreter: a chain of hashtables 12:41
On read, walk the chain looking for the key, and throw an exception if you can't find it
Assignment just binds to the top hashtable on the stack
It's pretty straightforward in that setting. But in a context like QAST, which likes to know about names up-front, it's a bit more tricky 12:42
12:46 pmurias left 12:49 Alina-malina joined, Alina-malina left, Alina-malina joined
timotimo i don't understand why jimmyz toldme to "please ignore ':'"? maybe he means using a version without binding? 12:50
yoleaux 10:09Z <JimmyZ> timotimo: perl6 --profile -e 'my int $n := 0; loop { $n := $n +1; if $n == 10000000 { exit; } }' has 10000001 Scalar Allocations, It will be nice to avoid it
10:09Z <JimmyZ> timotimo: please igonre ':'
10:16Z <JimmyZ> timotimo: and I see Exclusive Time on exit is 18446744073709.23ms ..
12:58 spider-mario joined
masak arnsholt: ties into what I felt was frustrating when I did some Python (3) last week: simple typos or thinkos resulting in "variable not found" errors were delayed until that line of code was executed at runtime. 12:59
12:59 JimmyZ joined
JimmyZ timotimo: yeah, binding to int won't work 12:59
13:00 JimmyZ is now known as Guest17214
timotimo ah, yes, binding to natives isn't supported yet 13:00
masak arnsholt: I can even see why they chose to go with that model (and similar design decisions pushing towards late binding and dynamism). but I'm so used to getting author-time errors about missing explicit lexical declarations... I guess I like that too much. 13:01
arnsholt Yeah, I can see the reasoning behind such a model. I just think it trades a small up-front cost for much larger downstream costs, and is thus actually a bad idea 13:03
masak agree.
FROGGS timotimo: he said "please ignore ':'", which probably means that the original code had assignment instead of binding
timotimo yeah
masak the lexical model is *beautiful*, both in the speed optimizations it permits and in the closure semantics it results in. 13:04
13:04 vti left
masak apparently closures were discovered by Steele and Sussman as they were investigating the Actors model! 13:06
timotimo ah, the scalar "allocations" are actually a bug 13:08
caused by my recent code 13:09
since the code uses getlexstatic_o, it gets the Scalar object, but getlexstatic_o doesn't always allocate
so what i do is i check if the object returned from it is at the very end of the nursery, which would mean "was just allocated" regularly
but in the case of this particular benchmark, it means "is the last thing that was allocated" 13:10
and that loop allocates nothing at all in the heap, so it'll count the same scalar object as allocated over and over again
Guest17214 and Exclusive Time :P
timotimo that's probably having recorded an entry time, but not an exit time
as the profiler dump code gets run as part of THE_END more or less 13:11
13:11 Guest17214 is now known as JimmyZ_
timotimo i'm wondering why the code isn't eliminating the $_ lexical 13:12
13:12 Ven joined
timotimo i replaced exit with last in order to get a proper profile 13:12
Ven o/, #perl6! 13:13
timotimo however, getting the static lexical "last" in that case is done as const_s lits(last) + getlexstatic_o
i'm not entirely sure what getlexstatic does exactly and if we have a direct rather than indirect version of that 13:14
but that could certainly help performance
oh, except
that BB only ever gets executed once. so never mind
masak \o Ven
Ven masak: I havn't backlogged yet, did you ping me at some point with some witty and interesting stuff? 13:15
timotimo weird. i seem to recall having put the optimize_can_op back in place, but here i see a const_s + can_s
Ven
.oO( like... A BLOG POST )
timotimo um ... Teh F? the jitlog shows "append label: 6", then a bunch of PHIs, then "append label: 6" and "append label: 6" 13:17
i'm *pretty* sure a label should only appear once, rather than three times
13:17 zakharyas joined 13:23 brrt joined
masak Ven: witty and interesting stuff like blog post still in our future time cone. :/ 13:24
Ven masak: too bad :). I secretly hoped the blog post you talked about before yesterday (/ yesterday) was there :P 13:25
brrt masak - never say ahead of time you're going to blog :-)
timotimo m) 13:26
i forgot to remove the break; before the call to optimize_can_op
masak brrt: point. 13:27
timotimo brrt: what do you say about the label situation i've mentioned up there? 13:31
brrt i'll be reading
backlogging, really
timotimo :)
our code-gen for a loop { } is using the while op and feeding it a QAST::Var(lexical "True") 13:32
13:32 ssqq joined
timotimo that's hilarious 13:32
even after speshing, that means we're looking up the contents of "True" for each iteration, guarding for concreteness, deconting, unboxing its int value and checking that against 0 13:33
m: my \True = 0; loop { say "lol" }; say "hahaha" 13:38
13:38 zakharyas left
camelia rakudo-moar e1aa07: OUTPUT«hahaha␤» 13:38
13:38 ssqq left
timotimo m: my \True = class :: { method Bool { (1, 1, 1, 1, 1, 0).pick } }; loop { say "lol" }; say "tadaa" 13:38
camelia rakudo-moar e1aa07: OUTPUT«lol␤lol␤lol␤lol␤lol␤lol␤tadaa␤»
13:39 ptc_p6 left, zakharyas joined
masak o.O 13:39
nice job, Houdini.
13:39 FROGGS left
masak I'd call it a bug if the urge to call it insane wasn't so much stronger. 13:40
Ven holy hell.
masak arguably, `loop` doesn't have anything to do with `True`.
timotimo yeah
arnsholt Nice one! 13:41
timotimo i changed it to emit IVal.new( :value(1) ) instead of the lexical True
a benchmark is about to come up.
masak are we *sure* we're not able to redefine 1 ? :P
Ven "you deserve what you get" 13:42
There was a trend in the JS ecosystem, before: wrap your code with `(function (undefined) { ... })()`
so that, if somebody redefined `undefined` (because it was possible), you'd still have a sane value! 13:43
I've yet to stop arguing about how stupid this is :)
timotimo time perl6 -e 'my int $n = 0; loop { $n = $n +1; if $n == 1_000_000_000 { last; } }'
11.96user 0.02system 0:12.01elapsed 99%CPU (0avgtext+0avgdata 123408maxresident)k
time perl6 -e 'my int $n = 0; loop { $n = $n +1; if $n == 1_000_000_000 { last; } }'
42.15user 0.02system 0:42.20elapsed 99%CPU (0avgtext+0avgdata 123348maxresident)k 13:44
guess which one is which
masak timotimo: yipes.
dalek kudo/nom: cd9001e | (Timo Paulssen)++ | src/Perl6/Actions.nqp:
loop { } doesn't need to inspect lexical "True" each iteration
 13:45
arnsholt That's a pretty nice speed-up!
timotimo++
timotimo i wouldn't call it "speed-up" for such a tight loop
but it's a nice overhead-reduction
Ven timotimo-- timotimo++ 13:46
arnsholt Troo
timotimo Ven: what did i deserve the -- for this time? 13:47
is it just your personal animousity against me?
Ven timotimo: you ++'d yourself :P
but I still wanted to ++ because you did something nice
timotimo oh
i ++'d myself by committing?
Ven oooh
timotimo++ # I'm blind, I read arnsholt's ++ as yours. My baad
timotimo fair enough 13:48
you karma-nazi :)
karmmar nazi?
krammer nazi.
masak Ven has personal animosity? I hope there's treatment for that.
Ven timotimo was mean to me for a week... That's merely payback :P
timotimo oh, without a u
masak .oO( there's no "you" in "animosity" ) 13:49
timotimo Ven: you were asking for it! :P
masak: no u!
Ven you made me install your module!
you monster.
masak timotimo: no, u!
13:50 kaare_ left 13:54 erkan joined, erkan left, erkan joined, guru joined, ptc_p6 joined, kaare_ joined, guru is now known as ajr_
timotimo is working on a fix to prevent the same object being counted as "allocated" infinitely many times in a row 14:03
"up to infinitely times" i should say 14:04
ah, yes, that seems to put things right.
masak m: say ^Inf 14:05
camelia rakudo-moar e1aa07: OUTPUT«0..^Inf␤»
masak m: say (^Inf).elems
camelia rakudo-moar e1aa07: OUTPUT«Inf␤»
timotimo done.
masak tries to look worried about the fact that the range excludes infinity, and thus does not quite go up to it 14:06
masak fails
timotimo masak: the thing is, you're also including the 0
so the count is actually Inf, not Inf - 1
you'd have to write 0^..^Inf for that
masak m: say 0 ^..^ Inf # neko operator
camelia rakudo-moar e1aa07: OUTPUT«0^..^Inf␤»
timotimo :D
need more manekineko in perl6 14:07
Ven m: sub infix:<(◕‿◕)>($a, $b) { $a + $b }; 3 (◕‿◕) 5 14:08
14:08 Celelibi left
camelia ( no output ) 14:08
Ven m: sub infix:<(◕‿◕)>($a, $b) { $a + $b }; say 3 (◕‿◕) 5 14:09
camelia rakudo-moar e1aa07: OUTPUT«8␤»
Ven why would you not love perl6?
14:11 Celelibi joined 14:15 hoelzro left 14:16 hoelzro joined
timotimo too lazy to get up and go to the desktop, hit the return key to wake it up, so i can launch a full benchmark run ... 14:17
i probably dread the possible realization that the nice 4x increase for the microbenchmark above probably translates into a 0.001% improvement in any use of loop { } that is more complex
s/possible/likely/ 14:18
14:19 sqirrel__ left, JimmyZ_ is now known as JimmyZ 14:20 sorear left, sorear joined
brrt it might be more given that - if JITTed - this might become a much simpler loop than it was 14:23
with much better branch prediction
timotimo would removing gotos that point to the immediately following instruction be a win at all? 14:25
brrt yes 14:26
albeit a small one
timotimo i bet branch prediction is very good at dealing with "goto" instructions
colomon hopes that question was rhetorical.
14:27 zakharyas left
brrt well, i would never bet on technology i don't fully understand 14:27
or that's not really true
timotimo colomon: i could imagine that the instruction decoding stage of the pipeline would just immediately drop goto statements on the floor if they are absolutely useless
brrt maybe 'fully' and 'never' need attenuating
Rounin We're still a bit behind on "Why does anything exist in the first place" question 14:28
timotimo "i would never fully bet on technology i don't fully understand" :)
Rounin I mean, a JIT without a universe would be pretty novel
Or without electromagnetism
14:28 pecastro left
colomon timotimo: but even if it did drop it, taking the time to drop it is still an issue. maybe a small one, but still an issue. 14:29
timotimo colomon: it's probably not as big an issue as removing a lexical lookup, or throwing out a redundant "set" instruction
colomon sure
brrt i build my next JIT on top of the weak nuclear force
timotimo brrt: then you'll get annihilated by the talented engineers at google who build their next javascript jit on top of the strong nuclear force ... 14:31
14:31 Ven left
timotimo hmm, what about building an anti-JIT? 14:31
brrt a BIT - backwards in time
timotimo :D
with a BTTF deoptimization engine
(and don't forget to compensate for register flux) 14:32
14:32 zakharyas joined
timotimo did i ever mention the "solipsism + 1" philosophy i (claim that i) developed? 14:33
basically, it's the belief that the universe exists completely without an intelligence to experience it 14:34
Rounin So your claim implies that you are not intelligent 14:35
timotimo hm. actually, no microbenchmark even has a loop { } in it
Rounin Do you suffer from that Coutard delusion or whatever it's called
timotimo hm, i think "an intelligence" isn't the word i meant
"a consciousness" perhaps? 14:36
Rounin But then you're not conscious
timotimo is that the one where a person believes they are already dead?
Rounin Who is making the claim?
Yeah haha :D
timotimo i remember that
it's not a claim, it's a philosophy :P
JimmyZ hmm, the last version is slower than the exit version, though the exit version have wrong time 14:37
I mean Exclusive Time :P 14:38
timotimo hm, could be
14:40 pecastro joined
timotimo ... i thought frame handlers were super cheap and there should not be a noticable overhead to having a last inside a loop like that? 14:41
brrt frame handlers are cheap, yes 14:42
JimmyZ except when they are not 14:43
brrt is not ready to get into discussions of 'consciousness' today
timotimo :D
brrt hang on, frame handlers are not free iirc 14:45
Ulti timotimo if I want to try and run your Cairo shooter on OSX do you have any tips for deps? 14:46
JimmyZ yeah, cheap doesn't mean free ;)
Ulti I have GTK and Cairo, but am having troubles with X server stuff and dbus
timotimo Ulti: give up right now. 14:47
brrt in the JIT, entering a frame handler means storing your bytecode position
timotimo Ulti: you can only find more pain
Ulti: but i've been working on an SDL2 port that works without GTK
[Coke] now getting malloc errors in rakudo-p
timotimo and without GTK you don't need X, because cairo is x-independent
brrt so that's relatively costly compared to not having it 14:48
timotimo brrt: did you read the label thing i found?
brrt no, not yet
overbusy :-) can you point me to it?
[Coke] S05-interpolation/regex-in-variable.rakudo.parrot..perl6-p(29944,0x7fff7987b310) malloc: *** error for object 0x7fcf1b9baad0: incorrect checksum for freed object - object was probably modified after being freed.
rurban: ^^ 14:49
timotimo brrt: let me try to reproduce it so you can look for yourself :)
[Coke] also timtoady, since that's a regex, and I know he just touched NQP.
also happening in S05-metasyntax/litvar.t. 14:50
Ulti timotimo got rid of the X11 fail, now it just segfaults moar :D
timotimo gist.github.com/timo/911901053290122928bb
Ulti: i'm not doing error checking very well
brrt odd... hmm 14:51
14:51 anaeem___ left
brrt i have my suspicions why this is so, but note that it doesn't 'really matter' much :-) 14:51
it is just inefficient
timotimo brrt: refresh and see more
how can emitting a label multiple times be correct?
oh, wait 14:52
are these just gotos?
brrt they're all emitted at the same place
no
they're labels
timotimo oh, because PHI have zero length
brrt basically, they do nothing but note to dynasm 'here's a special place'
and probably they come from ending/starting the frame handler labels
and several other ways you can get multiple labels 14:53
timotimo OK
that makes sense
brrt :-)
14:54 raiph joined
Ulti hmmm both cairo and gtk are working on their own for simple things like drawing a square and opening a blank window... 14:55
neat
and the only issue I had was not reading actionable output from brew when dbus was installed 14:56
14:57 sorear left
timotimo oh? 14:58
well, ven had a lot more trouble
14:58 sorear joined
[Coke] if uniq is now deprecated, need updates to the spectest suite. 14:59
timotimo damn. perl6-bench outputs the commits in american-date-order :\
14:59 kaleem left 15:00 virtualsue left
dalek k-simple: c9a432f | Matt++ | lib/GTK/Simple.pm6:
Deprecation of renamed method more -> emit

Haven't tested this.
Saw 1 call to deprecated code during execution.
  ================================================================================
Method more (from Supply) called at:
   lib/GTK/Simple.pm6, line 542
Deprecated since v2014.10, will be removed with release v2015.10! Please use emit instead.
  --------------------------------------------------------------------------------
Please contact the author to have these calls to deprecated code adapted, so that this message will disappear!		 15:01
15:02 [Sno] left
timotimo ah, yes. thank you! 15:03
15:03 zakharyas left
Ulti np 15:06
15:07 bjz_ joined, bjz left, KCL_ left
Ulti wistfully looks at WebKitGTK+ ^___^ 15:07
farabi could be a desktop app then 15:08
the amount of GTK and WebKit specific stuff you have to have working looks kind of crazy though
timotimo :\ 15:09
Ulti would be a lot easier to just automatically open a browser tab
timotimo might be easier to use the Chrome Embedding Framework, CEF
though i suspect that may be C++
15:13 KCL_ joined
nine If you're not tied to one browser engine, just open a browser for your app. The pain of embedding is just not worth it. 15:16
timotimo i agree 15:18
i recently tried to use "Brackets" or what it's called 15:19
i didn't get it to work at all.
JimmyZ uses Brackets too 15:20
and it's nice Javascript project 15:21
timotimo JimmyZ: would you mind telling me how to get it to run? 15:22
JimmyZ I download the .deb package
timotimo oh
i have fedora %)
15:23 telex left
JimmyZ
.oO( then only don't have .rpm)
 15:23
*doesn't
15:24 telex joined 15:25 KCL joined
rindolf JimmyZ: JavaScript is spelled with a capital "S". 15:26
timotimo .o( Java'S Crap, T! ) 15:27
.o( i pity de foolz who done Java )
hoelzro timotimo++ # github commentary
15:28 KCL_ left
JimmyZ As far as Perl 6 is faster enough, we can use Perl 6 write a new backend for rakudo, just like the java one :P 15:29
15:29 Ven joined
Ven masak: I'm not sure I linked it before. github.com/jashkenas/coffeescript/pull/3171 15:29
timotimo i'm glad you agree to my comment
JimmyZ: feel free to make perl6 fast enough so that we can bootstrap rakudo without nqp :)
JimmyZ :) 15:30
Ven .tell masak I'm not sure I linked it before. github.com/jashkenas/coffeescript/pull/3171 15:31
yoleaux Ven: I'll pass your message to masak.
masak computer, messages.
yoleaux 15:31Z <Ven> masak: I'm not sure I linked it before. github.com/jashkenas/coffeescript/pull/3171
masak Ven: interesting.
Ven Yeah. The discussion is a little bit interesting as well 15:32
masak aye.
timotimo brrt, JimmyZ, reducing the size of jitted code is always good for instruction cache reasons; so i suppose throwing out unnecessary gotos would be worth at least a tiny little something
Ven masak: in this case, they went with the "no ast sugar" approach. uncool :P 15:34
brrt timotimo: i agree 15:35
what do you have in mind
masak Ven: well, it's one approach.
Ven: substitute from the outside.
Ven masak: hence the linking :)
masak Ven: it *does* solve the {{{ }}} issue :P
Ven masak: have you read more of 50 years of lisp?
masak Ven: no, but I've ordered it.
Ven: at some point I will read the web page you linked, too. 15:36
Ven ordered and received homotopy type theory, but is still reading Chapter 0 by Aluffi for now... :)
masak Ven: patience. already later this week, I'll be more available.
Ven Well; I read a bit of HoTT already – and cried a lot.
masak cried?
Ven yes :)
timotimo brrt: was "what do you have in mind" aimed at me?
brrt yes 15:39
:-)
timotimo ah
just throwing out gotos at the end of BBs that have only a single successor anyway
hm ... actually
i think i may have to only drop gotos at the end of BBs if their single successor is also the linear_next bb 15:40
brrt seems ok yes 15:49
JimmyZ Good night
15:49 JimmyZ left
masak night, 卓明亮 15:50
15:50 erkan left 15:51 Ven left 15:52 erkan joined, erkan left, erkan joined 15:55 ajr_ left 15:57 guru joined, guru is now known as Guest36895, Guest36895 is now known as ajr_ 16:16 kaleem joined
timotimo brrt: the "big loop" benchmark that i set as the title of the that gist builds 478418 bytes of jitted bytecode. does that seem like a lot? 16:16
16:16 jluis left
brrt hmmm... it's a lot, yes, but not extraordinary i think 16:17
timotimo for 284 frames 16:18
brrt hmmm
and how many opcodes? 16:19
no, i don't think it is extraordinary
if anything it's extraordinary on the rakudo level
16:19 virtualsue joined 16:21 chenryn joined
timotimo m: 778 append_ins: <wval> 16:23
camelia rakudo-moar cd9001: OUTPUT«===SORRY!=== Error while compiling /tmp/opk813zHje␤Two terms in a row␤at /tmp/opk813zHje:1␤------> 778 ⏏append_ins: <wval>␤ expecting any of:␤ postfix␤ infix stopper␤ infix or met…»
timotimo 826 append_ins: <goto>
2124 append_ins: <set>
oops.
those are the top 3
rurban I fixed my append performance problems in potion by overallocating by 3. 16:25
16:25 erkan left
TimToady *3 or +3 ? 16:25
16:25 mr-foobar joined
rurban +3 was enough. I used for the 3 AST kids 16:26
16:26 acrussell joined
rurban But the number os append calls would be the same. 16:26
os=>of
TimToady I think p5 grows overextends arrays 20% or so, iirc 16:27
rurban in your case you could benchmark if worthwhile
was a huge win for me
TimToady unless it determined that you were shifting and pushing the same array, and then it would make it much larger to avoid lots of bulk copydowns 16:28
16:29 brrt left
TimToady but also a lot of numbers were tuned to the powers-of-two malloc that p5 would supply if you asked for it 16:30
16:32 fhelmberger left 16:33 TuxCM left
rurban not even +3, only mod 3. In blocks of multiples of 3. 16:34
b2gills I'm writing a post, what would you call this feature: 1,2,4 ... 32
timotimo m: say 1, 2, 4 ... 33 16:35
camelia rakudo-moar cd9001: OUTPUT«1 2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 131072 262144 524288 1048576 2097152 4194304 8388608 16777216 33554432 67108864 134217728 268435456 536870912 1073741824 2147483648 4294967296 8589934592 17179869184 34359738368 68719476736…»
timotimo that wasn't what i wanted :)
m: say 1, 99, 33 ... *
camelia rakudo-moar cd9001: OUTPUT«===SORRY!===␤Cannot assign to a readonly variable or a value␤»
timotimo oh, i remember now, that recently broke
TimToady b2gills: I dunno, deduced is the word used in the specs 16:37
so maybe 'deduced sequences'
b2gills I think I can work with that
timotimo i wanted to tickle that expression out of camelia 16:38
dalek rl6-roast-data: 76824d6 | coke++ | / (4 files):
today (automated commit)
 16:40
16:40 TuxCM joined
masak b2gills: it's a geometric sequence in this case. arithmetic and geometric sequences are the two we deduce. 16:41
b2gills I was going for more of a general name of the feature 16:42
16:43 denis_boyun left
rurban m: describe 1, 2, 4 ... 33 16:43
camelia rakudo-moar cd9001: OUTPUT«===SORRY!=== Error while compiling /tmp/qanQ3f0Rh0␤Undeclared routine:␤ describe used at line 1␤␤»
flussence "DWIM"? 16:45
TimToady I suppose deduced sequences are knowably monotonic, so we could detect end-point crossing, but if the user supplies the iterator, we can't know that easily 16:46
so there's no way to eliminate the surprise, but we could push it back a little 16:47
m: say 1, -2, 4 ... * 16:48
camelia rakudo-moar cd9001: OUTPUT«1 -2 4 -8 16 -32 64 -128 256 -512 1024 -2048 4096 -8192 16384 -32768 65536 -131072 262144 -524288 1048576 -2097152 4194304 -8388608 16777216 -33554432 67108864 -134217728 268435456 -536870912 1073741824 -2147483648 4294967296 -8589934592 17179869184 -34359…»
TimToady or knowably non-monotonic, except in abs 16:49
however, warning at compile time for a sequence known not to terminate would be...quite a trick, if it were a huge sequence 16:50
unless we have an easy formula for "Is this number in this sequence?" that doesn't involve calculating the whole sequence 16:51
and how much work do you put into that if you can't solve the general case, anyway?
I guess arithmetic termination is just modular, and the geometric case would tend to diverge pretty quickly for powers significantly over 1 16:53
colomon remembers that we changed to the current sequence rules because the old more DWIMy rules were a complete mess to implement… 16:54
TimToady the old dwimmy rules were trying to work for user-supplied functions as well, which doesn't 16:55
I guess for the geometric you want to look at representations in the base of the multiplier and see if the endpoint is representable with a single 1 bit followed by all zeros 16:58
so in the binary case, 33 is two set bits, not one
one could do something in logs to approximate the endpoint and see what matches, I suppose 16:59
17:00 ajr_ left
TimToady if there were a way of guaranteed rational logs for exactness, that would help 17:01
but I think that's probably asking too much of math :)
17:02 guru joined, guru is now known as Guest1875, Guest1875 is now known as ajr_ 17:03 kaleem left
timotimo today, i'd love to try fuzz testing out 17:05
since i've never done this before for any language or program ... tips please? :)
also: 17:06
m: say "a", "aa" ... *
camelia rakudo-moar cd9001: OUTPUT«a aa ab ac ad ae af ag ah ai aj ak al am an ao ap aq ar as at au av aw ax ay az ba bb bc bd be bf bg bh bi bj bk bl bm bn bo bp bq br bs bt bu bv bw bx by bz ca cb cc cd ce cf cg ch ci cj ck cl cm cn co cp cq cr cs ct cu cv cw cx cy cz da db dc dd de df dg…»
timotimo not what i mean ;(
m: say ("a", "b" ... "profit").elems
TimToady m: say [\~] 'a' xx * 17:07
17:07 ajr_ left
camelia rakudo-moar cd9001: OUTPUT«(timeout)» 17:07
rakudo-moar cd9001: OUTPUT«a aa aaa aaaa aaaaa aaaaaa aaaaaaa aaaaaaaa aaaaaaaaa aaaaaaaaaa aaaaaaaaaaa aaaaaaaaaaaa aaaaaaaaaaaaa aaaaaaaaaaaaaa aaaaaaaaaaaaaaa aaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaa aaaa…»
TimToady mine's shorter anyway :)
timotimo aye. not bad
17:08 ajr joined
TimToady m: say [\~]('a' xx *) ... 'aaaaaaaaa' 17:08
camelia rakudo-moar cd9001: OUTPUT«a aa aaa aaaa aaaaa aaaaaa aaaaaaa aaaaaaaa aaaaaaaaa␤»
17:08 ajr left
timotimo ooooooooh 17:09
17:10 ajr joined, ajr left 17:11 chenryn left, ajr joined, ajr left 17:15 ajr joined, ajr is now known as Guest80237 17:16 Guest80237 is now known as ajr_ 17:18 ajr_ left
masak m: say 'a', * ~ 'a' ... * 17:18
camelia rakudo-moar cd9001: OUTPUT«a aa aaa aaaa aaaaa aaaaaa aaaaaaa aaaaaaaa aaaaaaaaa aaaaaaaaaa aaaaaaaaaaa aaaaaaaaaaaa aaaaaaaaaaaaa aaaaaaaaaaaaaa aaaaaaaaaaaaaaa aaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaa aaaa…»
17:19 guru joined
TimToady m: say 'a' X~ 1..* 17:19
camelia rakudo-moar cd9001: OUTPUT«a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37 a38 a39 a40 a41 a42 a43 a44 a45 a46 a47 a48 a49 a50 a51 a52 a53 a54 a55 a56 a57 a58 a59 a60 a61 a62 a63 a64 a65 a66 a…»
TimToady m: say 'a' Xx 1..*
camelia rakudo-moar cd9001: OUTPUT«a aa aaa aaaa aaaaa aaaaaa aaaaaaa aaaaaaaa aaaaaaaaa aaaaaaaaaa aaaaaaaaaaa aaaaaaaaaaaa aaaaaaaaaaaaa aaaaaaaaaaaaaa aaaaaaaaaaaaaaa aaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaa aaaa…»
17:19 guru is now known as ajr_
timotimo that's cute 17:20
m: say 'a1' ... 'a99'
camelia rakudo-moar cd9001: OUTPUT«a1 a2 a3 a4 a5 a6 a7 a8 a9 b0 b1 b2 b3 b4 b5 b6 b7 b8 b9 c0 c1 c2 c3 c4 c5 c6 c7 c8 c9 d0 d1 d2 d3 d4 d5 d6 d7 d8 d9 e0 e1 e2 e3 e4 e5 e6 e7 e8 e9 f0 f1 f2 f3 f4 f5 f6 f7 f8 f9 g0 g1 g2 g3 g4 g5 g6 g7 g8 g9 h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 i0 i1 i2 i3 i4 i5 i…»
timotimo oh, hehe.
TimToady m: say 'a00' ... 'a99' 17:21
camelia rakudo-moar cd9001: OUTPUT«a00 a01 a02 a03 a04 a05 a06 a07 a08 a09 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37 a38 a39 a40 a41 a42 a43 a44 a45 a46 a47 a48 a49 a50 a51 a52 a53 a54 a55 a56 a57 a58 a59 a60 a61 a62 a63 …»
masak m: say 'a' X~ 00 .. 99
camelia rakudo-moar cd9001: OUTPUT«Potential difficulties:␤ Leading 0 does not indicate octal in Perl 6; please use 0o0 if you mean that␤ at /tmp/Nj7_GSh6rM:1␤ ------> say 'a' X~ 00⏏ .. 99␤a0 a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a1…»
masak m: say 'a' X~ 0 .. 99
camelia rakudo-moar cd9001: OUTPUT«a0 a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37 a38 a39 a40 a41 a42 a43 a44 a45 a46 a47 a48 a49 a50 a51 a52 a53 a54 a55 a56 a57 a58 a59 a60 a61 a62 a63 a64 a65 a6…»
TimToady m: say 'a00' ... 'a77'
camelia rakudo-moar cd9001: OUTPUT«a00 a01 a02 a03 a04 a05 a06 a07 a08 a09 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37 a38 a39 a40 a41 a42 a43 a44 a45 a46 a47 a48 a49 a50 a51 a52 a53 a54 a55 a56 a57 a58 a59 a60 a61 a62 a63 …»
masak m: say 'a' X~ ^100
camelia rakudo-moar cd9001: OUTPUT«a0 a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37 a38 a39 a40 a41 a42 a43 a44 a45 a46 a47 a48 a49 a50 a51 a52 a53 a54 a55 a56 a57 a58 a59 a60 a61 a62 a63 a64 a65 a6…»
masak m: say 'a' X~ (^100)>>.fmt("%02d") 17:22
camelia rakudo-moar cd9001: OUTPUT«a00 a01 a02 a03 a04 a05 a06 a07 a08 a09 a10 a11 a12 a13 a14 a15 a16 a17 a18 a19 a20 a21 a22 a23 a24 a25 a26 a27 a28 a29 a30 a31 a32 a33 a34 a35 a36 a37 a38 a39 a40 a41 a42 a43 a44 a45 a46 a47 a48 a49 a50 a51 a52 a53 a54 a55 a56 a57 a58 a59 a60 a61 a62 a63 …»
17:24 KCL_ joined 17:27 KCL left, andrewm joined, andrewm left 17:28 kaleem joined 17:29 [Sno] joined, kaleem left 17:35 pmurias joined
masak vendethiel: looking at it again, I think it's a bit of a stretch to call github.com/jashkenas/coffeescript/pull/3171 "hygienic". 17:36
[Coke] reminder, only one perl grant request came in this month, please comment it if you have something to say: news.perlfoundation.org/2014/11/gra...dux-1.html (it's a p5 module that maybe we could steal some stuff from) 17:38
pmurias arnsholt: pastie.org/9745189
arnsholt: as the paste shows the variables are declared at *compile* time 17:39
arnsholt: it's that it's checked an initialisation time if they were set
17:43 cognominal joined
masak pmurias: only for the very specific case of 'write before read', though. 17:43
pmurias meant to say "it's that it's checked at runtime if they were set" 17:47
masak: what my paste was meant to illustrate is that 'if 1==0: x = 123' declares x 17:49
17:49 dakkar left
masak indeed. 17:50
in other words, the scoping is per function, not per block. just like `var` in JavaScript.
17:51 pecastro left 17:52 ajr_ left
pmurias masak: they reference manual does not consider an if statement a block 17:53
"The following are blocks: a module, a function body, and a class definition."
masak oh, that makes sense.
but the difference to JavaScript is in terminology, not in semantics. 17:54
pmurias with no explicit way to declare a variable that sort of makes sense
masak: so my idea is that instead of using a hash table as a custom python scope, snake should keep a initalized flag 17:56
with my assumption being that a keeping some native int flags would be much cheaper than doing hash access 17:58
18:03 perturbation joined 18:05 alexghacker joined 18:08 khisanth__ joined 18:10 KCL_ left, khisanth_ left 18:13 raiph left 18:14 raiph joined 18:16 kjs_ left 18:17 guru joined 18:18 guru is now known as Guest41035, Guest41035 is now known as ajr_
alexghacker m: sub d( $first is copy, $second ) { say $first === $second }; my $a = []; my $b = {}; d( $a, $a ); d( $b, $b ) 18:18
camelia rakudo-moar cd9001: OUTPUT«True␤True␤»
18:18 gfldex joined
alexghacker From S03, === is "For objects that are not value types, their identities are their values." 18:20
18:20 xinming left
jnthn "is copy" dosn't mean "clone the thing" 18:20
alexghacker jnthn, I know it doesn't do a deep clone
jnthn It means "I want it assigned into a new Scalar, which I can assign to in my sub"
It doesn't do *any* clone.
alexghacker ah
that's my problem
(and why my deep clone attempts have all failed)
jnthn Yeah. Not sure there's a cute big of sugar for it 18:21
*bit
alexghacker the is copy is binding to the scalar holding the ref, not the ref itself
18:21 xinming joined
jnthn Yes, that's the way to understand it. 18:21
alexghacker jnthn++
now, to be clear, if the proto read: sub d( %a is copy ), then that would create a new hash container, correct? 18:23
jnthn iirc, it is compiled as if you had written %a = the_incoming_param 18:24
So yeah, i'd be a new hash container and then assign the incoming one's contents into it.
uh, it'd :) 18:25
timotimo 478481 ← after
jnthn Being a hash container ain't legal here... :P
timotimo 478418 ← before
so ... the opposite of a win?
jnthn timotimo: Is bigger better or worse? And what is the number for? :)
timotimo sum of bytes of jitted frames
18:26 rindolf left
timotimo um 18:26
what the hell did i do
18:26 FROGGS joined
timotimo i used the same jitlog for both invocations and got different numbers? 18:27
ah, good
throwing away gotos at the end of BBs if the linear_next was the target of the goto anyway
m: say "only { 478481 R/ 473545 } as many bytes as it used to generate" 18:28
camelia rakudo-moar cd9001: OUTPUT«only 0.9896840 as many bytes as it used to generate␤»
jnthn Gee, nqp-j takes a while to build...
moritz if you're used to moar... :-)
timotimo m: say "(that means we saved { 478481 R- 473545 } bytes)"
camelia rakudo-moar cd9001: OUTPUT«(that means we saved -4936 bytes)␤»
timotimo oops
m: say "(that means we saved { 478481 - 473545 } bytes)" 18:29
camelia rakudo-moar cd9001: OUTPUT«(that means we saved 4936 bytes)␤»
timotimo ~4.8 kbytes less for the instruction cache to nom sounds good
moritz r the setting?
s/r/for/
m: say 4936 / 2**10
camelia rakudo-moar cd9001: OUTPUT«4.820313␤»
timotimo moritz: that's just a simple tight loop 18:30
perl6 -e 'my int $n = 0; loop { $n = $n +1; if $n == 1_00_000_000 { last; } }' 18:31
moritz: i'll check out what amount of jit output the setting generates 18:32
oh, great! 18:33
/bin/sh: line 1: 31655 Segmentation fault (core dumped) /home/timo/perl6/rakudo/../install/bin/nqp-m tools/build/gen-cat.nqp moar -f tools/build/moar_core_sources > src/gen/m-CORE.setting
moritz m: say 4936 / 478481 18:36
camelia rakudo-moar cd9001: OUTPUT«0.0103160␤»
timotimo huh. my code to strip off redundant gotos at the end of BBs seems to be wrong. and the other way i thought it could be right is apparently also wrong 18:37
jnthn Apparently, you reduced the amount of JIT output the setting makes... :P 18:38
timotimo i don't get that far 18:39
moritz by some 100% :-)
jnthn I'd have thought it's just "check if the goto target is the linear next"
timotimo yeah, me too 18:40
if (bb->last_ins->operands[0].ins_bb == bb->linear_next) {
jnthn Uh, you are checking is an unconditional goto instruction also, yes? 18:41
timotimo (and only if it's a goto)
yes
otherwise that could be mayhem :)
jnthn Are these occuring in the code we're originally getting?
timotimo but it's very probably hard to accidentally hit the memory address of the bb's linear next
jnthn Or as a result of optimizations?
timotimo probably as a result of an optimization
jnthn ah
timotimo in that case, it could be safer to put a little extra check in and not emit the goto in that case 18:42
jnthn Was gonna say, if they're in the original code, we can more easily just drop them while we build the spesh graph.
timotimo i should check that out.
jnthn Yeah. Though the code that does it is actually the longest function in the whole of the VM, I think... :)
timotimo ouch! 18:43
i'll look some other way in that case :P
jnthn :P
It doesn't mean it's inherently hard.
Just long.
It finds the BB boundaries 18:44
And creates instruction nodes.
The later creates BBs, and finally puts them into a CFG
jnthn needs le diner 18:45
bbiab
18:48 smls joined
smls m: my $x = (2, 4, 6).BagHash; $x<4>--; say $x<4>; say $x.perl; say $x.kxxv 18:48
camelia rakudo-moar cd9001: OUTPUT«0␤(2=>1,4=>1,6=>1).BagHash␤2 4 6␤»
smls ^^ bug? 18:49
18:50 rindolf joined
colomon seems like 18:50
m: my $x = (2, 4, 6).BagHash; $x<4> = 0; say $x<4>; say $x.perl; say $x.kxxv 18:52
camelia rakudo-moar cd9001: OUTPUT«0␤(2=>1,4=>1,6=>1).BagHash␤2 4 6␤»
FROGGS o/ 18:54
smls m: my $x = (2, 4, 6).SetHash; $x<4> = False; say $x<4>; say $x.perl; # same with SetHash... 18:56
camelia rakudo-moar cd9001: OUTPUT«False␤SetHash.new(2,4,6)␤»
smls or are elements not supposed to automatically disappear when their weight is set to 0/False ?
19:03 Sqirrel joined
colomon they are suppose to disappear 19:03
and even if they don't, their value shouldn't be unchanged 19:04
m: my $x = (2, 4, 6).SetHash; $x<1000> = True; say $x<4>; say $x.perl;
camelia rakudo-moar cd9001: OUTPUT«False␤SetHash.new(2,4,6,"1000")␤»
colomon ah
that's your problem, maybe?
m: my $x = (2, 4, 6).SetHash; $x{4} = False; say $x.perl;
smls oh, right
camelia rakudo-moar cd9001: OUTPUT«SetHash.new(2,6)␤»
smls colomon++ 19:05
colomon yeah, not a bug, just an awkward thing
smls nah its fine, thats how object hashes that check identity with .WHICH are supposed to work...
19:07 pmurias left
timotimo aye. 19:14
good catch, colomon
smls Maybe Perl 7 will allow us to write BagHash[Int()].new(2, 4, 6) to get a mutable bag that automatically coerces all inputs to Int... :) 19:16
or Perl 6.x 19:17
19:17 mullagain joined
TimToady std: my Int() %bh is BagHash = 2,4,6; 19:18
camelia std 76ccee1: OUTPUT«ok 00:00 140m␤»
smls "is" ? 19:19
TimToady yes, that's how you replace the default Hash with something else, in theory 19:20
timotimo if you don't do that, you'll end up with a regular hash instead
or you'd have to bind
smls cool
doesn't seem to work yet, though
TimToady don't think it implemented thoguh
smls btw, why are types not coercive by default?
They are in C++, aren't they?
timotimo that's a different thing 19:21
in C++, the decisions will be done at compile time, no? and only for subclasses 19:22
moritz types aren't generally coercive in C++
timotimo we can coerce from different types in p6
like Str to Int/Num/Bool/...
smls well thats only for things that are coded to expect/work with Cool
not part of the type system itself 19:23
moritz though maybe with a copy constructor you can emulate it
mullagain is rakudo the standard compiler thus far?
19:25 sqirrel__ joined
moritz mullagain: yes (though standard by emergence, not by definition) 19:25
smls moritz: Well, methods like .Str/.Int sort of are the Perl 6 version of copy constructors, aren't they? 19:26
wait, no, the opposite
mullagain perl6 is using pthreads right? 19:29
19:29 sqirrel__ left
TimToady perl6 is using a hybrid model 19:30
work is scheduled at a higher abstraction level and can be allocated to any of a number of real threads 19:31
mullagain ok thanks!
smls I like how Perl 6 lets you schedule a pice of code to be run on another thread 19:33
in Perl 5, you had to implement a custom thread pool for every use-case, because you could only send simple data to other threads, not code objects 19:34
so the thread pool that started the threads, had to know about and implement all necessary functionality in advance 19:36
masak vendethiel: also, the macro system is vulnerable to multiple evaluation when parameterized nodes have side effects (such as `$i++`). but then, most macro systems are, I think.
vendethiel masak: oh, it's pretty bad
19:36 PZt left
vendethiel masak: but it's ... trying :) 19:36
masak I'm not sure there's a good solution to that, except on a case-by-case basis inside of the macro.
mullagain does <key> replace {key}? 19:37
vendethiel mullagain: yes and no. key cannot be used to mean "key" anymore
mullagain: so you can use {"key"} or <key>.
19:37 kurahaupo1 joined
TimToady heh, "A New Jersey hiker, mulled to death by a bear in September..." 19:37
mullagain ah ok, so the < is shorthand 19:38
masak m: sub key { say "bar" }; my %h = foo => "foo", bar => "bar"; say %h<key>; say %h{key}
camelia rakudo-moar cd9001: OUTPUT«(Any)␤bar␤(Any)␤»
masak m: sub key { "bar" }; my %h = key => "foo", bar => "bar"; say %h<key>; say %h{key}
camelia rakudo-moar cd9001: OUTPUT«foo␤bar␤»
masak mullagain: ^
TimToady hopes not to be mulled to death... 19:39
kurahaupo1 TimToady: slowly boiled to death?
TimToady I'll have to mull that over...
mullagain haha
TimToady anyway, that popped up in Google News a bit ago 19:40
19:42 PacoLinux joined, kurahaupo left, anaeem1 joined
mullagain so rakudo is written in java? 19:43
masak no, Rakudo is written in mostly Perl 6 and NQP. 19:45
some of the JVM backend stuff is written in Java.
19:45 PacoLinux left
mullagain i was just compiling and i saw stuff like this /home/mike/rakudo-nom/install/bin/nqp-j --target=jar --output=blib/Perl6/Compiler.jar --encoding=utf8 \ 19:45
19:48 user3 joined, jluis_ joined
hoelzro is there a QAST::Op for getting the current global context of the module being loaded? 19:48
moritz mullagain: rakudo supports three backends; oneof them is the JVM
hoelzro I think github.com/rakudo/rakudo/blob/nom/...d.nqp#L361 is missing a parameter to the current GLOBALish, and I'd like to see if inserting that fixes things 19:49
19:49 user3 left 19:50 dwarring joined 19:53 anaeem1 left
dwarring java interop question on SO - stackoverflow.com/questions/2715632...from-perl6 19:56
19:59 darutoko left
dalek p: d0a0d50 | jonathan++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/ (2 files):
Don't allocate longlit array every NFA run.
 20:00
20:06 raiph left, raiph joined
smls m: class A { has $.n; method WHICH { "a" } }; say set A.new(n => 1), A.new(n => 2), A.new(n => 3) 20:11
camelia rakudo-moar cd9001: OUTPUT«set(A.new(n => 3))␤»
smls :P
20:19 kurahaupo1 left 20:22 kurahaupo joined 20:25 vti joined 20:26 khisanth__ is now known as Khisanth 20:30 Mouq joined
smls m: my $a = [[1, 2], [3, 4]]; my $b = $a.deepmap(*.clone).Array; $a[0][0] = 7; say $b.perl 20:33
camelia rakudo-moar cd9001: OUTPUT«[[1, 2], [3, 4]]␤»
smls ^^ Is this the best way to deep-clone an array?
20:33 kurahaupo left
moritz it's not deep enough, I fear 20:35
smls ? 20:36
moritz m: my $a = [{ a => [1]}]; my $b = $a.deepmap(*.clone).Array; $a[0]<a>[0] = 42; say $b.perl
camelia rakudo-moar cd9001: OUTPUT«[{"a" => [1]}]␤»
moritz seems I'm wrong
Mouq m: my $a = [{ a => b => [1]}]; my $b = $a.deepmap(*.clone).Array; $a[0]<a><b>[0] = 42; say $b.perl 20:40
camelia rakudo-moar cd9001: OUTPUT«[{"a" => "b" => [42]}]␤»
smls so it doesn't descend into a Pair, because it doesnit count as a map'able collection? 20:47
makes sense I guess
20:54 Mso150 joined 20:56 prime left
moritz the problem is really that it hits .clone eventually 20:58
and .clone isn't deep
21:03 Mso150 left
smls m: my $a = [{ a => b => [1]}]; my $b = EVAL $a.perl; $a[0]<a><b>[0] = 42; say $b.perl 21:05
camelia rakudo-moar cd9001: OUTPUT«[{"a" => "b" => [1]}]␤»
smls :)
21:05 anaeem1_ joined, Mso150 joined 21:06 kurahaupo joined 21:08 prime joined 21:11 rindolf left 21:12 raiph left
timotimo o/ 21:17
21:23 perturbation left
dalek ast: ffc87fa | usev6++ | S03-sequence/basic.t:
Add second test for RT #78324
 21:26
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...l?id=78324
21:33 brrt joined 21:34 raiph joined 21:35 vti left 21:37 prime left
timotimo webpagetest tells me rakudo.org's front page could benefit from gzip transfer encoding (down to 18%) and giving caching meta-information to style.css, recaptcha.css, another style.css, rakudo-1001.png (interestingly from rakudoperl.org), wordpress.png and favicon.ico 21:38
tony-o btyler: i'm working on a refactor for http server async that is looking like it handles async (like the name implies) significantly better, what siege settings were you using? 21:39
btyler: gist.github.com/tony-o/3f7ac835659b0152d1eb
moritz timotimo: I haven't done anything on rakudo.org, because I know I can't fix it if I break it 21:40
masak TimToady: re github.com/perl6/specs/commit/f22e...b5308c5701 -- what's the rationale for character classes not participating in longest literal matching? 21:41
timotimo i understand
tony-o: yays!
i like hearing that
i wonder how we could implement farabi6 on top of HTTP::Server::Async if available (aka if MoarVM) and otherwise run it on HTTP::Server::Simple 21:42
tony-o farabi6 is that ide?
timotimo yes, it is
i think it could benefit from asynchronous features 21:43
tony-o i might try to port a templating system i wrote for p5
moritz masak: because in the end, everything comes down to either literals or character classes
tony-o not sure yet, though
timotimo we don't need templating for farabi :)
moritz masak: and if you equate them, there's no disambiguation anymore
21:44 telex left
timotimo with the example you gave me for taking over an existing connection i should be able to implement EventSource server-side, right? 21:44
tony-o: and how hard would it be to add keepalive to HTTP::Server::Async?
21:44 telex joined
tony-o timotimo: not really sure, it's a total rewrite of the Async.pm6 so give me a couple of days to get it to where the current one is. i don't imagine adding keep-alive should be too tough 21:45
timotimo OK :) 21:46
do you have a clue how hard HTTP2/SPDY is?
21:47 alexghacker left
tony-o i don't, i was reading libmicrohttpd source earlier today though 21:47
21:48 prime joined 21:49 alexghacker joined
tony-o i need to read more about spdy 21:49
timotimo spdy is being deprecated by http2 anyway 21:50
it'll mostly be interesting to find out how http2 became the way it is
masak moritz: that's not true. both literals and character classes have width. zero-width assertions don't, so they (for example) could be the thing that halts longest-lit. ditto quantifiers and/or capturing groups. 21:57
21:57 jluis_ left
moritz masak: well, let me rephrase 21:57
masak: in a successful parse, every character of a string is consumed either by a literal or by a char class 21:58
masak: so it's just a very natural approach to say that a char consumed by a literal is more specific than one consumed by a char class, thus it should get precedence 21:59
masak no argument there.
I guess what I'm asking about is what "force" made TimToady decide that `<[abc]>` differs in semantics from `[ a | b | c ]` 22:00
("force" in the "forces acting on the problem" sense.)
moritz not being TimToady, I can only speculate :-) 22:01
masak I can totally see how `\w` is not considered part of a literal prefix. I have a harder time making that argument with `<[abc]>`.
FROGGS masak: when you parse something, you usually have keywords, and these usually should be preferred over \w+ and friends
masak well, see what I just wrote. 22:02
moritz masak: what about a..z ?
there seems to be a spectrum here
FROGGS examples with one characters things does not help here
masak moritz: I'm aware that it's a slippery slope. especially since `\w` could also go inside `<[ ]>`
FROGGS: disagree. `<[abc]>` is (in some sense) *literal*, but `\w` is not. 22:03
moritz literals, enumerations, ranges, properties, negated ranges, negated enumerations
masak moritz: right.
moritz: which is why one possible answer I can imagine from TimToady is "I figured it was just easier to draw the line earlier".
moritz and it would feel equally weird to say that <[abc]> behaves differently from <[a..c]>
masak: also char class vs. non-char-class is very clear line 22:04
masak right, that's what I mean.
by "easier".
moritz masak: except that you wrote "earlier" :-)
FROGGS masak: show me a parser where you want to treat <[abc]> equally to keywords
masak FROGGS: /<[Ff]>alse/ 22:05
FROGGS keywords are more important in the way they also define an order of the chars
moritz FROGGS: also there's FORTRAN, where variables staring with certain letter ranges get certain default types; I can well imagine the use of a char class there. But overall, I agree with your point 22:06
FROGGS: it's very practical, whereas we have been discussing things more from a purity PoV
FROGGS masak: that slightly feels like a "misuse" of cclasses :o)
22:07 kaare_ left
masak FROGGS: oh, so you would prefer /[F|f]alse/ ? 22:08
FROGGS letting LLTM also treat cclasses like literals just to compress your grammar might not be the best choice
masak FROGGS: then will you be the one to explain to the newbie why they are different? :)
FROGGS masak: I think I prefer /False | false/ 22:09
masak FROGGS: that is also an unsatisfactory cop-out.
TimToady: will the zany `G<*n|enesis>` syntax participate in longest lit? :)
FROGGS everything is about tradeoffs... and I'd like to have the line between cclasses and literals where it is now 22:10
I hope TimToady can argue better than I can :o) 22:11
masak me too :)
alexghacker so I've done a bunch of work trying (and failing) to implement an naive but general deepclone, but while reading S32 I think I found that there's one more-or-less built in
m: my $h = { a => 1, b => [ 2, 3, { c => 4, d => 5, f => { g => -1 } } ] }; my $i = $h.deepmap( { $_ } ); $i{'b'}[2]{'e'} = 6; $i{'b'}.unshift(0); $i{'e'} = 7; say $h; say $i
camelia rakudo-moar cd9001: OUTPUT«"a" => 1, "b" => [2, 3, {"f" => {"g" => -1}, "d" => 5, "c" => 4}]␤"a" => 1, "b" => [0, 2, 3, {"e" => 6, "f" => {"g" => -1}, "d" => 5, "c" => 4}], "e" => 7␤»
alexghacker .deepmap( { $_ } ) appears to do the trick perfectly 22:12
FROGGS I guess one problem is that <[abc]> is a too simple view on these kind of cclasses
masak alexghacker: for a mix of arrays, hashes, *and* arbitrary objects? 22:13
alexghacker masak: not for arbitrary objects
alexghacker wasn't looking for that to begin with
masak ok.
alexghacker although if the object were Iterable, then in theory deepmap would recurse into it 22:16
(exactly what that would mean for an "exotic" Iterable object that was a proxy for a database is left unsaid) 22:17
22:17 anaeem1_ left, ptc_p6 left
alexghacker cloning objects properly isn't something that can be solved generally anyway. the object has to cooperate. 22:18
raydiak good afternoon #perl6 22:19
22:20 denis_boyun joined 22:25 Mso150 left 22:27 denis_boyun left
raydiak was thinking about junctions last night...why don't we want people introspecting them? everyone says "use a set", but sets don't autothread which is the reason people try to use junctions in that way in the first place 22:30
so then I was thinking about a different type of autothreading container class, but without knowing what the attitude towards junctions is about, I can't exactly evaluate the viability of my ideas
22:32 bartolin left, bartolin joined 22:33 brrt left 22:36 rurban left, ajr_ left
tony-o how do i determine if IO::Socket::Async.listen was successful? or die doesn't seem to work even if the listen failed 22:39
22:40 gfldex left
TimToady raydiak: you're basically asking for the proposed Each type 22:40
22:41 FROGGS left
jnthn tony-o: Same as with any Supply, you subscribe to the quit event 22:41
(Given as a named argument to .tap)
TimToady on the alternation vs cclass thing, we already had to handle alternations for longest literals, so the only question is whether to extend it further. Going to any character class is a big jump, at least from an implementation point of view, so that's where I drew the line 22:42
22:43 dadada joined
dadada hi 22:43
raydiak TimToady: thanks, I can stop thinking about it then
dadada so parrot is dead, yes?
TimToady parrot: say "Hi, I'm dead." 22:44
p: say "Hi, I'm dead."
camelia rakudo-parrot cd9001: OUTPUT«Hi, I'm dead.␤»
dadada ok, thought so
Mouq I think that's TimToady++'s cryptic way of saying "no"... 22:45
:P
jnthn
.oO( Was the choice ofg "yes yes yes" as a nick when asking that question deliberate? :) )
*of
Anyways, sleep time here...should have more Perl 6 time again soon, to continue the native array work. :) 22:46
o/
TimToady you think that's cyrillic?
jnthn TimToady: No, just translit'd :)
I'm so tired I see things in all kinds of ways by this point :P
TimToady maybe it's just a dadadaist
jnthn :) 22:47
&
dadada if you really want to know my nick is inspired by this www.youtube.com/watch?v=vMZcXQCYwG0
22:49 alexghacker left
tony-o jnthn: github.com/tony-o/perl6-http-serve...nc.pm6#L48 22:50
jnthn: i have that but never see it executed when i try to connect to the server and recv a 'connection refused' 22:51
22:53 alexghacker joined 23:06 woolfy joined 23:09 lizmat joined 23:13 kjs_ joined 23:16 spider-mario left
smls TimToady: If Set() is a coercer, why does it take arbitrarily many arguments? 23:17
m: say Set(2, 3, 'a'=>4) 23:18
camelia rakudo-moar cd9001: OUTPUT«set(2, 3, a)␤»
smls I'm getting the feeling that subroutine forms of coercers don't really carry their weight. 23:20
23:21 woolfy left
smls Little benefit for the added confusingness (e.g. they don't exactly reinfoce the message of the GLR separation of methods invocants vs subroutine args) 23:23
What is the reason for having them, other than to have Int() as direct replacement of P5's int()? 23:24
raydiak tony-o: sorry if this is too obvious, but if the server appears to listen without error, are you sure your requests are reaching it and not hitting a filter/firewall/whatever? 23:29
23:31 denis_boyun_ joined
raydiak tony-o: if it helps narrow it down, examples/echoserver works fine here on your dev branch 23:34
23:45 kjs_ left 23:51 Ugator left 23:54 adu joined
adu hey 23:54
raydiak hi adu
adu I have an algo question
how do I write a function that takes f(x, [a, b, c]) 23:55
and gives "<apply>" ~ a ~ "<apply>" ~ b ~ "<apply>" ~ c ~ x ~ "</apply></apply></apply>"
23:55 virtualsue left
raydiak m: sub foo ($x, @($a, $b, $c)) {.say for $x, $a, $b, $c}; foo(1, [2, 3, 4]) 23:57
camelia rakudo-moar cd9001: OUTPUT«1␤2␤3␤4␤»
adu um, what about a..z? 23:58
like maybe rv = x, then rv ~= "<apply>" ~ item ~ rv ~ "</apply>" 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!