»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
[Coke] how does one install SVG::Plot for rakudo? 00:03
[Coke] finds panda. 00:05
panda borked on current nom? 00:06
00:15 ch3ck3r left
flussence missing some regex stuff, iirc. Lots of stuff seems to currently... 00:17
[Coke] rakudo: class A { multi method a() { }}; use MONKEY_TYPING; augment class A { multi method a() { } } 00:23
p6eval rakudo 545638: ( no output ) 00:24
00:25 colomon joined
[Coke] rakudo: for 1,2 X <a b> { say "$^x $^y" } 00:25
p6eval rakudo 545638: OUTPUT«1 a␤1 b␤2 a␤2 b␤»
[Coke] for 1,2 X (<a b> X 'x') { say "$^x $^y" }
rakudo: for 1,2 X (<a b> X 'x') { say "$^x $^y" } 00:26
p6eval rakudo 545638: OUTPUT«1 a␤1 x␤1 b␤1 x␤2 a␤2 x␤2 b␤2 x␤»
00:28 colomon_ joined
[Coke] rakudo: 555 ~~!~~ 666 00:29
p6eval rakudo 545638: ( no output )
[Coke] rakudo: my $a=305+437+6; say ($a-44), " tickets that aren't just blocking on tests." 00:30
p6eval rakudo 545638: OUTPUT«704 tickets that aren't just blocking on tests.␤»
[Coke] rant: "LTA error message" is not, by itself, an actionable ticket. 00:31
00:31 colomon left, colomon_ is now known as colomon 00:33 Chillance left
colomon [Coke]: I know the thought on the LTA error message tickets is to remind us to be better. But obviously it's a pretty low priority thing! 00:34
[Coke] rakudo: @_.=[3] 00:38
p6eval rakudo 545638: OUTPUT«===SORRY!===␤Cannot declare placeholder parameter @^_ in the mainline at line 1, near ".=[3]"␤»
00:39 colomon left
[Coke] rakudo: say 1 % 0 00:41
p6eval rakudo 545638: OUTPUT«1␤»
00:41 replore joined
[Coke] std: class A {}; sub A {} 00:41
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Illegal redeclaration of routine 'A' (see line 1) at /tmp/1e7_lwVnRT line 1:␤------> class A {}; sub A⏏ {}␤Check failed␤FAILED 00:01 119m␤»
[Coke] rakudo: class A {}; sub A {}
p6eval rakudo 545638: ( no output )
[Coke] rakudo: say 16 < 5 < 66 00:42
p6eval rakudo 545638: OUTPUT«Bool::False␤»
[Coke] rakudo: say (16 < * < 66)(5)
p6eval rakudo 545638: OUTPUT«Bool::True␤»
00:45 whiteknight left 00:46 whiteknight joined
[Coke] rakudo: my $a = class { has $a; }; say $a.perl; 00:47
p6eval rakudo 545638: OUTPUT«<anon>␤»
[Coke] rakudo: my $a = class { has $a; }; say $a.new(a => 1).perl;
p6eval rakudo 545638: OUTPUT«<anon><-856690017062603138>␤»
00:48 whiteknight left
[Coke] rakudo: sub if() { '2' }; say if; 00:49
p6eval rakudo 545638: OUTPUT«␤»
benabik nom: say if;
p6eval nom 545638: OUTPUT«␤Could not find sub &if␤ in block <anon> at /tmp/eVdvJS7KRE:1␤ in <anon> at /tmp/eVdvJS7KRE:1␤␤»
[Coke] boy, some of these masak bugs are sticky.
rakudo: my %h = {"a" => "b"}; %h{"a"} = %h.delete("a"); say %h.perl; 00:50
p6eval rakudo 545638: OUTPUT«().hash␤»
diakopter std: while (4) ; # hopefully this is legal
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Missing block at /tmp/0gQBL4sLsC line 1:␤------> while (4) ⏏; # hopefully this is legal␤Parse failed␤FAILED 00:01 120m␤»
[Coke] rakudo: multi foo() { say "OH HAI" }; multi foo(Int $a?) {};multi foo(Str $a?) {}; foo 00:51
p6eval rakudo 545638: OUTPUT«Ambiguous dispatch to multi 'foo'. Ambiguous candidates had signatures:␤:()␤:(Int $a)␤:(Str $a)␤␤ in sub foo at /tmp/exSGgscNNw:1␤ in block <anon> at /tmp/exSGgscNNw:1␤ in <anon> at /tmp/exSGgscNNw:1␤␤»
[Coke] rakudo: die "A"; CATCH { say "OH HAI" }; CATCH { say "OH NOES" } 00:52
p6eval rakudo 545638: OUTPUT«===SORRY!===␤only one CATCH block allowed at line 1, near ""␤»
[Coke] rakudo: die "A"; CONTROL { say "OH HAI" }; CONTROL { say "OH NOES" }
p6eval rakudo 545638: OUTPUT«===SORRY!===␤only one CONTROL block allowed at line 1, near ""␤»
00:56 S11001001 left
[Coke] rakudo: 1 .. 2 .. 3 00:56
p6eval rakudo 545638: ( no output )
[Coke] rakudo: my @a = <one two>; @a[-1] = 'zero'; @a.perl 01:00
p6eval rakudo 545638: OUTPUT«Cannot use negative index -1 on Array␤ in method throw at src/gen/CORE.setting:5980␤ in method <anon> at src/gen/CORE.setting:6074␤ in <anon> at src/gen/Metamodel.pm:2085␤ in find_method_fallback at src/gen/Metamodel.pm:2083␤ in find_method at src/gen/Metamode…
[Coke] std: my @a = <one two>; @a[-1] = 'zero'; @a.perl 01:01
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Unsupported use of [-1] subscript to access from end of array; in Perl 6 please use [*-1] at /tmp/HGjlVLl5ks line 1:␤------> my @a = <one two>; @a[-1]⏏ = 'zero'; @a.perl␤Parse failed␤FAILED 00:01 123m␤»
[Coke] is ::T valid anymore?
01:05 Tene left, djanatyn left
TimToady why not? 01:09
[Coke] Just wondering if I could close some tickets, is all. 01:10
TimToady: there are several [spec] tickets that probably need you to poke at them.
rakudo: multi foo() { nextsame }; foo() 01:11
p6eval rakudo 545638: ( no output )
TimToady don't doubt it
[Coke] rakudo: ::() 01:12
p6eval rakudo 545638: OUTPUT«===SORRY!===␤Cannot look up empty name␤»
[Coke] std: ::() 01:13
p6eval std 20ae3bd: OUTPUT«ok 00:01 119m␤»
[Coke] is rakudo ok there?
TimToady sure
rakudo: ::(()) 01:15
p6eval rakudo 545638: ( no output )
01:15 ggoebel left
benabik A variable named '()'? 01:16
TimToady should warn about undefined 01:17
[Coke] woot. there are now 747 tickets in the queue.
benabik nom: say ().WHAT
[Coke] there are 47 tickets that needtests.
p6eval nom 545638: OUTPUT«Parcel()␤»
01:18 alester_ joined
TimToady rakudo: ::(Nil) 01:19
p6eval rakudo 545638: ( no output )
TimToady rakudo: ::(Any)
p6eval rakudo 545638: OUTPUT«Use of uninitialized value in string context␤»
TimToady rakudo: ().Str 01:20
p6eval rakudo 545638: ( no output )
TimToady rakudo: say ().Str
p6eval rakudo 545638: OUTPUT«␤»
TimToady rakudo: say Nil.Str 01:21
p6eval rakudo 545638: OUTPUT«␤»
TimToady std: sub if() { '2' }; say if; 01:24
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Unsupported use of bare 'say'; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument at /tmp/L2jtRs4j7l line 1:␤------> sub if() { '2' }; say⏏ if;␤Confused at /tmp/L2jtRs4j7l line 1:␤-----…
TimToady std: sub if() { '2' }; say if();
p6eval std 20ae3bd: OUTPUT«ok 00:01 122m␤»
01:47 wolfman2000 joined 02:44 donri left 02:51 ponbiki joined 02:56 colomon joined 03:10 mishin joined 03:12 kaleem joined 03:15 colomon left 03:22 envi_ joined
sorear good * #perl6 03:25
03:29 colomon joined 03:35 scorpil left, frew joined, frew left 03:39 colomon left
mishin Hi perl6 03:42
i try to perl Configure.pl --gen-parrot under cygwin and get error gist.github.com/1263566 03:46
but i think i need site.icu-project.org/ 03:50
sorear Configure picked a compiler that doesn't exist. Wierd. 03:52
04:04 tokuhir__ joined 04:07 tokuhiro_ left, envi_ left 04:36 lestrrat left 04:41 satyavvd joined 04:48 [Coke] left
moritz good morning 04:54
05:00 alester_ left
sorear hi moritz 05:03
05:03 mishin left 05:12 molaf joined 05:15 lestrrat joined 05:22 SHODAN joined 05:29 dual joined
moritz nom: sub foo() { die "OH NOEZ" }␤ foo 05:54
p6eval nom 545638: OUTPUT«OH NOEZ␤ in sub foo at /tmp/2I7guEZtFD:1␤ in block <anon> at /tmp/2I7guEZtFD:2␤ in <anon> at /tmp/2I7guEZtFD:1␤␤»
moritz nom: sub foo() { die "OH NOEZ" }; foo 05:55
p6eval nom 545638: OUTPUT«OH NOEZ␤ in sub foo at /tmp/BnNlgTp7PK:1␤ in block <anon> at /tmp/BnNlgTp7PK:1␤ in <anon> at /tmp/BnNlgTp7PK:1␤␤»
06:01 wolfman2000 left
japhb perl6: (0 but True).say 06:10
p6eval rakudo 545638: OUTPUT«No applicable candidates found to dispatch to for 'infix:<but>'. Available candidates are:␤:(Mu $obj, Positional @roles)␤:(Mu $obj, Mu $role)␤␤ in sub infix:<but> at src/gen/CORE.setting:7364␤ in block <anon> at /tmp/PtemMnfgKk:1␤ in <anon> at /tmp/PtemMnfgKk:1␤…
..pugs: OUTPUT«*** Cannot cast from VBool True to Pugs.AST.Types.VCode (VCode)␤ at Prelude.pm line 541, column 5-16␤»
..niecza v10-36-g241a2d8: OUTPUT«␤Unhandled Exception: RoleApply with superclasses NYI␤ at /home/p6eval/niecza/lib/CORE.setting line 0 (CORE infix:<but> @ 0) ␤ at /tmp/pKkgy0gIkN line 1 (MAIN mainline @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2115 (CORE C1022_ANON @ 2) ␤ at …
06:10 tokuhir__ left
japhb OK, are those just three different ways to say 'NYI', or am I missing something obvious? 06:11
perl6: (0 but Bool::True).say 06:13
p6eval rakudo 545638: OUTPUT«No applicable candidates found to dispatch to for 'infix:<but>'. Available candidates are:␤:(Mu $obj, Positional @roles)␤:(Mu $obj, Mu $role)␤␤ in sub infix:<but> at src/gen/CORE.setting:7364␤ in block <anon> at /tmp/T_XPTD3evU:1␤ in <anon> at /tmp/T_XPTD3evU:1␤…
..niecza v10-36-g241a2d8: OUTPUT«␤Unhandled Exception: RoleApply with superclasses NYI␤ at /home/p6eval/niecza/lib/CORE.setting line 0 (CORE infix:<but> @ 0) ␤ at /tmp/OgHvadB67x line 1 (MAIN mainline @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2115 (CORE C1022_ANON @ 2) ␤ at …
..pugs: OUTPUT«*** Cannot cast from VBool True to Pugs.AST.Types.VCode (VCode)␤ at Prelude.pm line 541, column 5-16␤»
sorear japhb: 2 NYI, 1 oldspec
also:
japhb sorear, OK, thanks
sorear b: (0 but True).say 06:14
p6eval b 1b7dd1: OUTPUT«0␤»
japhb OOC, why is b: not included in perl6: ? Seems at least as useful a comparison point as pugs: ...
sorear ask moritz 06:15
japhb moritz, ^^ :-)
phenny, ask jnthn Can you make '$value but True' and '$value but False' work without too much trouble? This is coming up when trying to implement MAIN as per spec. 06:17
phenny japhb: I'll pass that on when jnthn is around.
sorear japhb: try writing it longform
rakudo: say (0 but role { method Bool { True } })
p6eval rakudo 545638: OUTPUT«0␤»
sorear rakudo: say ?(0 but role { method Bool { True } })
p6eval rakudo 545638: OUTPUT«Bool::True␤»
japhb sorear, ah ... OK, that will work for now. 06:18
sorear per spec, this is exactly the same as 0 but True
japhb phenny, tell jnthn sorear++ has found a workaround. A bit uglier, but the long form seems to work.
phenny japhb: I'll pass that on when jnthn is around.
06:19 mtk left 06:21 gnude joined 06:27 donri joined 06:32 tokuhiro_ joined 06:33 mtk joined 06:41 gnude left 06:44 mberends joined
dalek ecza: fce895a | mberends++ | examples/gtk-sierpinski.pl:
[examples/gtk-sierpinski.pl] enable window resizing
 06:46
06:47 GlitchMr joined
sorear hello GlitchMr 06:48
GlitchMr Hi, sorear
mberends hi sorear: Niecza had a very good at amsterdam.pm.org last night imo. The only disappointment mentioned was startup speed, and I promised the members that the serialize branch would fix that :) 06:51
*very good showing # I should read before pressing Enter
we looked at Gtk stuff and also the p5 integration examples. 06:52
sorear should try to find a pm someday to understand what mberends is talking about better 06:58
06:59 mj41 joined
sorear but not this week, too few tuits 06:59
and on that note, sleep
mberends good night
GlitchMr std: aaa{} 07:04
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'aaa' used at line 1␤Check failed␤FAILED 00:01 119m␤»
GlitchMr std: sub aaa{Because $aaa}
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Variable $aaa is not predeclared at /tmp/K5uHUxQbLn line 1:␤------> sub aaa{Because ⏏$aaa}␤Undeclared name:␤ 'Because' used at line 1␤Check failed␤FAILED 00:01 120m␤»
07:09 colomon joined 07:31 mishin joined 07:43 mj41 left 07:47 colomon left 07:48 baest joined 07:55 mj41 joined
dalek ast: ef8462e | moritz++ | S10-packages/basic.t:
test class redeclaration of a package
 07:55
moritz nom: sub f($x:) { say $x }; f(3) 07:58
p6eval nom 545638: OUTPUT«Lexical 'self' not found␤ in sub f at /tmp/VVBtLcDj2G:1␤ in block <anon> at /tmp/VVBtLcDj2G:1␤ in <anon> at /tmp/VVBtLcDj2G:1␤␤»
moritz nom: sub f($x:) { say $x }; 3.&f
p6eval nom 545638: OUTPUT«Lexical 'self' not found␤ in sub f at /tmp/ShMWDE7gCI:1␤ in method dispatch:<var> at src/gen/CORE.setting:689␤ in block <anon> at /tmp/ShMWDE7gCI:1␤ in <anon> at /tmp/ShMWDE7gCI:1␤␤»
moritz nom: sub f($x:) { say $x };
p6eval nom 545638: ( no output )
07:58 amkrankruleuen joined
moritz nom: my method f($x:) { say $x }; 1.&f 07:58
p6eval nom 545638: OUTPUT«1␤»
dalek ast: 044a4a6 | moritz++ | S12-subset/subtypes.t:
test for RT #71820, failed subset type check error includes name of subset
 08:05
moritz nom: say (1 Z 2).perl 08:06
p6eval nom 545638: OUTPUT«((1, 2),).list␤»
moritz nom: say (1 Z 2).[0].WHAT
p6eval nom 545638: OUTPUT«Parcel()␤»
moritz nom: say (1 Z 2).elems
p6eval nom 545638: OUTPUT«1␤»
dalek ast: 64bc53d | moritz++ | S03-metaops/zip.t:
test that infix Z returns a list of parcels (RT #75818
 08:09
08:11 amkrankruleuen left, amkrankruleuen joined 08:12 tokuhiro_ left
dalek ast: bb44b6c | moritz++ | S10-packages/joined-namespaces.t:
new test file for joined namespaces
 08:14
kudo/nom: 3d67ff2 | moritz++ | lib/Test.pm:
[Test] do not count dying will Null PMC access as a success
 08:16
kudo/nom: 840ad74 | moritz++ | t/spectest.data:
run another test file
08:27 masak joined
masak *yawn* good morning, #p6ople 08:27
phenny masak: 04 Oct 23:09Z <diakopter> ask masak how many primes in the first 10^9 primes are primes when represented in a number system other than decimal, but interpreted as decimal, or vice versa?
masak: 04 Oct 23:11Z <diakopter> ask masak ... or in any number system represented as another number system but interpreted as the original number system?
masak: 04 Oct 23:15Z <diakopter> ask masak ... e.g. the prime X represented in base-X is 1, etc. that'll get you started.
masak: 04 Oct 23:16Z <diakopter> ask masak er, scratch that last line ;)
masak :P 08:28
a kind of "number system mojibake" problem... 08:29
moritz I don't quite understand the problem or question 08:30
masak in the case of interpreting in a fewer-digit number system, there will also be lots of "illegal" numbers. just stating the obvious here, really.
moritz there's a known number of primes below 1e9
and then you can think about number of ways to represent them
masak hm, indeed.
moritz or is it about strings that are primes when interpreted in two different ways? 08:31
masak oh, it appears so.
moritz like '13'.base(10) and '13'.base(8) both being prime, even if it not the same?
nom: say '13'.base(8) 08:32
p6eval nom 545638: OUTPUT«Method 'base' not found for invocant of class 'Str'␤ in block <anon> at /tmp/UAf5bHITZN:1␤ in <anon> at /tmp/UAf5bHITZN:1␤␤»
moritz erm, wrong way around
nom: say :8('13')
p6eval nom 545638: OUTPUT«11␤»
08:32 dakkar joined
masak :) 08:32
I'm on irssi on screen on feather today, by the way.
<jlaire> sounds very ad-hoc 08:35
this is the feeling I get whenever "interpreted in base X" shows up.
moritz funny thing is, you can even use changing bases 08:39
08:39 mj41 left
moritz ie a different base for each digit 08:39
and in some of those bases, you can represent pi and e and so on as very regular numbers 08:40
pi/2 is 1.111111... in the base (1, 1/3, 2/5, 3/7, 4/9 ...) 08:42
nom: my @base := 1..* Z 1, 3, 5 ... *; say @base[^6].perl 08:44
p6eval nom 545638: OUTPUT«((1, 1), (2, 3), (3, 5), (4, 7), (5, 9), (6, 11))␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say @base[^6].perl
p6eval nom 545638: OUTPUT«(1/1, 2/3, 3/5, 4/7, 5/9, 6/11)␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say [+] 2 X* @base[^20] 08:45
p6eval nom 545638: OUTPUT«22.4796732103645␤»
moritz erm, right, I need to build the running product of these bases
nom: my @base := 1..* Z/ 1, 3, 5 ... *; say [+] 2 X* [ @base[^20]
p6eval nom 545638: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<[ ]>, couldn't find final ']' at line 1␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say [+] 2 X* [\*] @base[^20]
p6eval nom 545638: OUTPUT«5.14157671400225␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say [+] -2, 2 X* [\*] @base[^30] 08:46
p6eval nom 545638: OUTPUT«0␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say [+] 2 X* [\*] @base[^30]
p6eval nom 545638: OUTPUT«0.115608257536946␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say [+] 2 X* [\*] @base[^25] 08:47
p6eval nom 545638: OUTPUT«1.14158113096343␤»
moritz nom: my @base := 1..* Z/ 1, 3, 5 ... *; say pi - [+] 2 X* [\*] @base[^25]
p6eval nom 545638: OUTPUT«2.00001151903657␤»
moritz ... modulo the first digit, it appears :-)
08:48 packetknife left
masak nom: my @base := 1..* Z/ 1, 3, 5 ... *; say (pi - [+] 2 X* [\*] @base[^25]) % 1 08:49
p6eval nom 545638: OUTPUT«1.151903657437e-05␤»
masak close enough :)
moritz a better way to use this identity are so-called "spigot algorithms" 08:50
dalek kudo/nom: fc304e3 | moritz++ | src/core/Exception.pm:
report errors to $*ERR
 08:51
ast: 18c3485 | moritz++ | S0 (3 files):
tests for RT #77056 and #76280
 08:52
08:52 woosley joined 09:01 daxim joined
masak re rt.perl.org/rt3//Public/Bug/Displa...l?id=73502 09:06
the current rakudo error message is plain wrong.
and AFAIAC, there's nothing wrong with the code as such.
moritz nom: @_.= 3 09:07
p6eval nom 840ad7: OUTPUT«===SORRY!===␤Cannot declare placeholder parameter @^_ in the mainline at line 1, near ".= 3"␤»
moritz well, it's @_ that's being declared, not @^_
masak right.
moritz but apart from that... I agree
masak in fact, nothing's being declared as such.
moritz @_ *is* a placeholder var in Perl 6
masak nom: my @_ 09:08
p6eval nom 840ad7: OUTPUT«===SORRY!===␤Cannot declare placeholder parameter @^_ in the mainline at line 1, near ""␤»
09:08 [Coke] joined
moritz masak: would you prefer 'Cannot *use* placeholder parameter"? 09:09
09:11 fngraph joined, fngraph left
masak yes, I think that would help a lot. 09:14
because placeholder parameters aren't declared, really. that's why we have them.
dalek kudo/nom: b7af9ff | moritz++ | src/Perl6/Actions.pm:
improve error message, masak++
 09:16
moritz masak: it's nearly as easy to change such things as to whine about them :-) 09:17
s/whine/complain/
masak good to know.
moritz the misreporting of @_ as @^_ might be less trivial to fix, but shouldn't be too hard either 09:18
masak although (1) at this point, I have complaining down to a fine art, and (2) there's still a bit of a threshold to start making commits while doing $dayjob. 09:19
09:22 Vlavv_ joined 09:25 Vlavv` left 09:30 orafu left, replore left 09:32 orafu joined 09:35 GlitchMr left 09:42 wamba joined 09:43 woosley left
moritz nom: say -Inf cmp -Inf 09:43
p6eval nom 840ad7: OUTPUT«0␤»
moritz nom: say -Inf cmp '-Inf' 09:44
p6eval nom 840ad7: OUTPUT«-1␤»
moritz nom: say '-Inf'.Numeric
p6eval nom 840ad7: OUTPUT«-Inf␤»
moritz teh fun
don't worry, I'm about to fix it
t/spec/S32-io/IO-Socket-INET.t is sooo slow 09:53
10:01 pothos left 10:02 pothos_ joined, pothos_ is now known as pothos 10:06 envi_ joined 10:15 ggoebel joined 10:16 mishin left
dalek kudo/nom: bbe823d | moritz++ | src/Perl6/Actions.pm:
avoid @_ being reported as @^_

It seems easier to pass the full variable name to add_placeholder_parameter than to reconstruct it from sigil, identifier and the twigil		 10:17
10:17 sirmacik left, sirmacik joined 10:39 f00li5h left 10:45 colomon joined 11:07 tokuhiro_ joined
dalek kudo/nom: 4f19b79 | Coke++ | t/spectest.data:
run unfudged test
 11:12
masak [Coke]: LTA-class or not, if a ticket is too vague to have an observable, please flag it as problematic. 11:14
[Coke]: I know I sometimes make assumptions when posting tickets. I try not to.
tadzik good morning, herd
in what time is #phasers today? 11:17
11:17 colomon left
masak tadzik: it was 17:30 UTC last week. 11:20
tadzik hmm 11:23
so it's in the middle of my lectures /
:/ 11:24
if I count correctly
bbl
masak those are late classes. 11:25
tadzik aye 11:26
whole day free, and one lecture 18-20
it'd better be an interesting one
moritz nom: say <foo bar baz>.min 11:31
p6eval nom bbe823: OUTPUT«bar␤»
moritz the current implementation in nom does that by setting a tracker to +Inf 11:32
then using $tracker cmp $elem on each, and relies on that facth that +Inf cmp's higher than any other element
nom: say 'f' cmp 'I' 11:33
p6eval nom bbe823: OUTPUT«1␤»
moritz nom: say 'foo' cmp 'Inf'
p6eval nom bbe823: OUTPUT«1␤»
moritz nom: say 'foo' cmp '-Inf'
p6eval nom bbe823: OUTPUT«1␤»
moritz ... and it coerces both arguments to .Numeric first, to see if one of them is Inf or -Inf 11:34
which goes horribly wrong in some cases
nom: class A { has $.s handles <Str> }; say (A.new(:s<a>), A.new<:s<b>).min 11:35
p6eval nom bbe823: OUTPUT«Method 'at_key' not found for invocant of class 'A'␤ in method postcircumfix:<{ }> at src/gen/CORE.setting:1018␤ in block <anon> at /tmp/Dps1k06ACJ:1␤ in <anon> at /tmp/Dps1k06ACJ:1␤»
11:35 satyavvd left
moritz nom: class A { has $.s handles <Str> }; say (A.new(:s<a>), A.new(:s<b>).min 11:35
p6eval nom bbe823: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 1␤»
moritz nom: class A { has $.s handles <Str> }; say (A.new(:s<a>), A.new(:s<b>)).min
p6eval nom bbe823: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:608␤ in sub infix:<==> at src/gen/CORE.setting:1993␤ in sub infix:<cmp> at src/gen/CORE.setting:1036␤ in method reify at…
moritz so that happens because cmp looks if $a == Inf etc. 11:36
masak moritz: maybe Inf should really be a Num, but in all other cases (Int, Str, etc) it stands in for something meaning "always wins in a comparison".
well, not always. not against Inf, I guess.
11:37 Psyche^ joined
moritz masak: but then we need a better to check if something is an Inf 11:37
nom: class A { has $.s handles <Str> }; say A.new(:s<a>) == Inf 11:38
p6eval nom bbe823: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:608␤ in sub infix:<==> at src/gen/CORE.setting:1993␤ in block <anon> at /tmp/PkejLY3KPA:1␤ in <anon> at /tmp/PkejLY3KPA:…
moritz nom: class A { has $.s handles <Str> }; say A.new(:s<a>).?Numeric == Inf
p6eval nom bbe823: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:608␤ in method dispatch:<.?> at src/gen/CORE.setting:719␤ in block <anon> at /tmp/5htm48hpin:1␤ in <anon> at /tmp/5htm48…
11:38 kaleem left
moritz somehow I liked the old .? semantics better 11:38
masak huh? .?Numeric fails with an error? and that's the "new semantics"? 11:39
moritz masak: yes, because there a Mu.Numeric, which it finds 11:40
11:41 Patterner left, Psyche^ is now known as Patterner, GlitchMr joined
moritz tl;dr I have no idea how to fix infix cmp 11:42
11:46 replore joined
masak it's the same fundamental issue as with sorting: we basically have no idea how to factor types into it all. 11:47
nom: say Mu.Numeric 11:48
p6eval nom bbe823: OUTPUT«Use of uninitialized value in numeric context␤0␤»
masak nom: say Mu.new.Numeric
p6eval nom bbe823: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:608␤ in block <anon> at /tmp/jRZGlh0zm5:1␤ in <anon> at /tmp/jRZGlh0zm5:1␤»
masak huh.
moritz that's intentional
masak it's a Liskov smell, at least. 11:49
moritz you basically never want to numify a type object to anything else than 0
flussence wow, I left tryrakudo on its own for a few weeks... and it's still running
moritz flussence: does it run on nom already? 11:50
flussence er, lemme check (I think so)
oh, nope
2011.02-4-g14319fc
shouldn't be too hard to fix...
moritz probably just involves uploading a src/gen/CORE.setting.pir to feather3 11:51
flussence oh wait, it's slightly more involved since the safe mode stuff changed
moritz it got easier :-)
but yes, it still needs adaption 11:52
11:54 bluescreen10 joined 11:55 baest left 11:56 baest joined
dalek kudo/nom: ff78b5b | moritz++ | tools/build/Makefile.in:
clean up CORE.setting
 11:58
masak moritz++ # tireless improvements
12:06 preflex left
masak biggest beef I have with irssi so far, after using it for half a day, is that the input buffer is separate from the channel discussion. 12:07
12:07 preflex joined
masak which is totally irrational, since it has no bearing at all on what can and can't be done. it's just a presentational thing. 12:08
12:17 bluescreen10 left
masak today's mini-challenge: write a "template system" in as little code as possible. 12:18
it should take as input a template and a bunch of key-value mappings to be replaced. 12:19
12:19 bluescreen10 joined
masak the exact format of the template and the way the key-value mappings are replaced are left open for easier golfability. 12:19
moritz so just interpolate variables, no ifs/loops/whatever? 12:20
masak right. 12:26
moritz masak: gist.github.com/1264326 12:34
note that with a bit fewer bells and whistles, that would be just %args{$0} 12:35
masak will look, will just post my variant first :) 12:36
nom: my $t = q[%foo% %bar% %foo% %baz%]; my %b = foo => 1, bar => 2, baz => 3; $t.=subst("%{.key}%", .value, :g) for %b.pairs; say $t
p6eval nom 4f19b7: OUTPUT«Method 'key' not found for invocant of class 'Any'␤ in block <anon> at /tmp/_FcaSjNVCT:1␤ in block <anon> at /tmp/_FcaSjNVCT:1␤ in method reify at src/gen/CORE.setting:3749␤ in method reify at src/gen/CORE.setting:3654␤ in method reify at src/gen/CORE.setting:365…
moritz the &filter is for injecting HTML escaping etc.
masak moritz: nice! 12:37
b: my $t = q[%foo% %bar% %foo% %baz%]; my %b = foo => 1, bar => 2, baz => 3; $t.=subst("%{.key}%", .value, :g) for %b.pairs; say $t
p6eval b 1b7dd1: OUTPUT«1 2 1 3␤»
masak what's wrong with nom here? 12:38
[Coke] moritz++ 738 tickets; 40 blocking on tests.
moritz I'm not sure it's actually wrong
maybe it is
masak: it doesn't give the block inside the string a $_
[Coke] calls that "under 700 ``real'' tickets, but needs a better word. 12:39
moritz nom: { .say }.()
p6eval nom 4f19b7: OUTPUT«Any()␤»
[Coke] feels dirty using fake ascii double quotes. ;)
moritz nom: $_ = 'foo'; { .say }.()
p6eval nom 4f19b7: OUTPUT«foo␤»
moritz nom: $_ = 'foo'; "{ .say }"
p6eval nom 4f19b7: OUTPUT«foo␤»
moritz huh
nom: $_ = 'foo'; "{ .say }" for $_
p6eval nom 4f19b7: OUTPUT«foo␤»
moritz b: my $t = q[%foo% %bar% %foo% %baz%]; my %b = foo => '%bar%', bar => '%foo%', baz => 2; $t.=subst("%{.key}%", .value, :g) for %b.pairs; say $t 12:40
p6eval b 1b7dd1: OUTPUT«%foo% %foo% %foo% 2␤»
moritz masak: are you sure you want that semantics for your template system?
:-) 12:41
the first rule of parsing is that you avoid multi-pass parsing.
... with exceptions :-) 12:42
[Coke] ... parse club. you should say "parse club."
moritz [Coke]: you're right :-)
[Coke] rakudo: class A { has $.a, $.b; }; A.new(a => 1, b => 10); 12:43
p6eval rakudo 4f19b7: OUTPUT«Null PMC access in find_method('b')␤ in <anon> at /tmp/1VU50j3jQh:1␤ in block <anon> at /tmp/1VU50j3jQh:1␤ in <anon> at /tmp/1VU50j3jQh:1␤»
dalek ecza: 23f1b2e | (Paweł Murias)++ | lib/ (2 files):
remove debugging statements
 12:44
12:44 pmurias joined 12:45 GlitchMr left
[Coke] does a niecza spectest and sees that things did not stay where he put them. 12:45
[Coke] suspects people are, like, adding new tests and stuff.
masak moritz: for sure, it's a tradeoff between minimalism and that weird multi-pass behavior. I should probably use .trans instead. 12:46
moritz [Coke]: I'm usually not very good at testing my spectest changes with niecza first 12:47
[Coke]: otoh I'm fine with cleaning up tests or rakudo when somebody adds tests for niecza 12:48
12:48 benabik left
masak b: my %b = foo => 1, bar => 2, baz => 3; say .key for %b.pairs 12:48
p6eval b 1b7dd1: OUTPUT«foo␤bar␤baz␤» 12:49
masak nom: my %b = foo => 1, bar => 2, baz => 3; say .key for %b.pairs
p6eval nom 4f19b7: OUTPUT«foo␤bar␤baz␤»
masak moritz: I still don't get why my mini-templating system works on b but not on nom.
moritz nom: my %b = foo => 1, bar => 2, baz => 3; say "{.key}" for %b.pairs
p6eval nom 4f19b7: OUTPUT«Method 'key' not found for invocant of class 'Any'␤ in block <anon> at /tmp/Jsj9_WhUJ_:1␤ in block <anon> at /tmp/Jsj9_WhUJ_:1␤ in method reify at src/gen/CORE.setting:3749␤ in method reify at src/gen/CORE.setting:3654␤ in method reify at src/gen/CORE.setting:365…
moritz b: my %b = foo => 1, bar => 2, baz => 3; say "{.key}" for %b.pairs
p6eval b 1b7dd1: OUTPUT«foo␤bar␤baz␤»
moritz nom: say "{.key}" given a => b
p6eval nom 4f19b7: OUTPUT«Could not find sub &b␤ in block <anon> at /tmp/c5YU2FpqXA:1␤ in <anon> at /tmp/c5YU2FpqXA:1␤»
moritz nom: say "{.key}" given a => 1
p6eval nom 4f19b7: OUTPUT«Method 'key' not found for invocant of class 'Any'␤ in block <anon> at /tmp/2uyJfcJMNu:1␤ in block <anon> at /tmp/2uyJfcJMNu:1␤ in block <anon> at /tmp/2uyJfcJMNu:1␤ in <anon> at /tmp/2uyJfcJMNu:1␤»
moritz b: say "{.key}" given a => 1 12:50
p6eval b 1b7dd1: OUTPUT«a␤»
moritz nom: $_ = a => 1; say "{.key}"
p6eval nom 4f19b7: OUTPUT«a␤»
masak odd, no?
moritz yes, seems like topicalization by 'given' and 'for' does something differently
12:51 pmurias left
moritz ... different than $_ assignment, that is 12:51
masak submits rakudobug
12:52 tokuhiro_ left 12:53 Sarten-X left
[Coke] there are many new failures in the spec test on niecza which make me think the new tests are wrong. 12:57
for example: S06-traits/is-rw.t
nok eval('foo(28)'), 'is rw requires a variable'; 12:58
this causes niecza to die with:
Unhandled Exception: Binding $a is rw in MAIN foo, cannot bind read-only value to is rw parameter
is the test wrong there, or niecza? 12:59
13:00 im2ee joined 13:01 Sarten-X joined 13:02 cosimo left
im2ee Hello! :) 13:09
Woodi hi im2ee 13:10
[Coke]: from what are listen std can answer such questions 13:11
std: eval('foo(28)')
p6eval std 20ae3bd: OUTPUT«ok 00:01 120m␤»
[Coke] Woodi: no, std is just a parser. 13:12
Woodi then no idea what you ask about :)
[Coke] doesn't help when you're trying to decide how an eval should fail.
phenny: tell sorear that when I fudge roast for niecza, he should definitely check to make sure he's happy with the new tests. 13:13
phenny [Coke]: I'll pass that on when sorear is around.
[Coke] there. buck passed.
Woodi [Coke]: then it looks like something like anti-wiki :)
13:14 cosimo joined 13:15 benabik joined 13:16 Tene joined
masak [Coke]: &eval doesn't catch excceptions any more, so to me it looks like the test is wrong, yes. 13:16
13:18 cosimo_ joined 13:19 tzhs joined
Woodi b: my $a = 1; say $a.undef.defined 13:24
p6eval b 1b7dd1: OUTPUT«Method 'undef' not found for invocant of class 'Int'␤ in main program body at line 22:/tmp/FTec92shi6␤»
Woodi nom: my $a = 1; say $a.undef.defined
p6eval nom ff78b5: OUTPUT«Method 'undef' not found for invocant of class 'Int'␤ in block <anon> at /tmp/WNUtMR_pYC:1␤ in <anon> at /tmp/WNUtMR_pYC:1␤»
Woodi std: my $a = 1; say $a.undef.defined 13:25
p6eval std 20ae3bd: OUTPUT«ok 00:01 122m␤»
Woodi std: my $a = 'a'; say $a.undef.defined
p6eval std 20ae3bd: OUTPUT«ok 00:01 122m␤»
Woodi b: my $a = 'a'; say $a.undef.defined
p6eval b 1b7dd1: OUTPUT«Method 'undef' not found for invocant of class 'Str'␤ in main program body at line 22:/tmp/lU3Rjb0ZET␤»
masak Woodi: STD won't tell you when a method doesn't exist. 13:26
it's just a parser.
Woodi b: my $a = 'a'; $a = Any; say $a.defined
p6eval b 1b7dd1: OUTPUT«Bool::False␤»
masak std: my $a = 1; $a.completely-made-up-method-name
p6eval std 20ae3bd: OUTPUT«ok 00:01 122m␤»
Woodi k 13:27
b: my $a = 'a'; undef $a; say $a.defined 13:28
p6eval b 1b7dd1: OUTPUT«===SORRY!===␤Unsupported use of undef as a verb; in Perl 6 please use undefine function or assignment of Nil at line 22, near " $a; say $"␤»
Woodi b: my $a = 'a'; undefine $a; say $a.defined
p6eval b 1b7dd1: OUTPUT«Bool::False␤»
Woodi tada
masak nom: my $a = 'a'; undefine $a; say $a.defined 13:29
p6eval nom ff78b5: OUTPUT«Bool::False␤»
Woodi b: my $a = 'a'; undefine $a; say $a.perl 13:30
p6eval b 1b7dd1: OUTPUT«Any␤»
Woodi sub a{ return undefine }; say a.perl 13:31
b: sub a{ return undefine }; say a.perl
p6eval b 1b7dd1: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in 'undefine' at line 606:src/gen/core.pm␤ in 'a' at line 22:/tmp/Vg60EQJ50P␤ in main program body at line 22:/tmp/Vg60EQJ50P␤»
Woodi nom: sub a{ return undefine }; say a.perl 13:32
p6eval nom ff78b5: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in sub undefine at src/gen/CORE.setting:7467␤ in sub a at /tmp/jIK5ew30wq:1␤ in block <anon> at /tmp/jIK5ew30wq:1␤ in <anon> at /tmp/jIK5ew30wq:1␤»
masak 'undefine' is a function, expecting 1 argument.
Woodi: you seem to assume it's a term.
Woodi it's not right for me
dalek ast: 2f1dec0 | Coke++ | S0 (10 files):
niecza fudging
Woodi perl5: sub a{ return undef }; print a() 13:33
[Coke] ok. if anyone updates those tests, I'd appreciate a retry in niecza. I guess I just need to auto-smoke after roast updates.
Woodi is any v5 daemon here ? :)
masak Woodi: 'undef' is both a term and a function in Perl 5. in Perl 6, the term is now Any (and other type objects), and 'undefine', respectively. 13:34
Woodi do not like this
masak s/, and/, and the function is/
Woodi 'return Any' looks strange...
masak Woodi: feel free to explain why you do not like it. I like it.
or 'return Str' if your function is supposed to return a Str. 13:35
or 'fail'.
or 'die "something went wrong"'
Woodi becouse i learned from Perl5 and 'return undef' was perfect, clear, etc
[Coke] Some things are going to change. Sorry. 13:36
Woodi but why change good things ??
to many mantras IMO
masak Woodi: you can't just come into #perl6 and ask "but why change good things ??" :P 13:38
it's our job to take an essentially good thing -- Perl 5 -- and change it for the better.
huf Woodi: also, in perl5 return undef was most probably a bug :) 13:39
masak these changes are often done with good reason. if the reason turned out to be less than good, we usually make a better change later.
huf: there's that, too.
Woodi: when would you do 'return undef' rather than 'return' in Perl 5? 13:40
Woodi return returns truth i think...
huf you think wrong 13:41
but i think in this case the best you can do is figure out how undefs work and why it was split into many things in p6 before pronouncing judgement
masak perl6: say defined foo; sub foo { return } # Woodi: observe 13:42
p6eval rakudo ff78b5, niecza v10-38-g23f1b2e: OUTPUT«Bool::False␤»
..pugs: OUTPUT«␤»
huf masak: a bare return is short form for what in p6? 13:43
in p5 it was return () iirc
masak Woodi: I'd be happy to explain why Perl 6 has two things where Perl 5 has only one. but as huf says, it's more urgent that you know about how undefs work in Perl 5. :)
huf: 'return Nil'
IIUC.
and Nil is short for a kind of undef () 13:44
Woodi huf: "RFC 28: Perl should stay Perl" and: "The essence of Perl is really context sensitivity, not just to syntactic context, but also to semantic, pragmatic, and cultural context."
huf masak: Nil is what? empty list in list ctx and some form of undef in item
?
Woodi: yes? 13:45
Woodi after all Perl6 is descendant of Perl5 and good things should stay
masak huf: yes, something very much like that.
huf you've just demonstrated that you dont understand how the supposed good think in p5 works
Woodi: how *exactly* do you want it to stay the same in p6 if you dont know what it is?
masak huf: it's an object representing "could've got a list of things here, but got nothing, not even an empty list"
huf masak: ah, Nil is emptylister than ()? heheh :D 13:46
snarkyboojum thank goodness for good error messages
std: undef
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Unsupported use of undef as a value; in Perl 6 please use something more specific:␤ Mu (the "most undefined" type object),␤ an undefined type object such as Int,␤ Nil as an empty list,␤ :!defined as a matcher,␤ Any:U as a type constrain…
masak snarkyboojum++
Woodi huf: how you think, why Perl is/was so popular ? :) becouse some clear things like "return undef" JUST WORKS
huf Woodi: but i'm telling you it's most likely a big in p5. 13:47
masak Woodi: as huf points out, you're making a bit of a fool of yourself by claiming that Perl 6 should be like Perl 5, and then demonstrating your lack of insight into how Perl 5 actually works.
[Coke] Woodi: there is always perl5, and you are, of course, welcome to keep using it if you don't want to learn more about six.
13:47 localhost left
Woodi allPpls: i think perl5 wasn't perfect and this is my reason to involwing in v6. but "what was worked will stay" - cannot find exact citation... 13:48
mux reminds me of yesterday, when someone who had broken his FreeBSD system for not properly following 3rd party apps update instructions, claimed that FreeBSD is broken, that versioned shared libs are too, and that I was an idiot for not believing him
13:49 localhost joined
masak Woodi: you keep assuming that 'undef' was working as it was. 13:49
PerlJam Woodi: "return undef" is often a code smell anyway
Woodi yes, masak, it was working perfect. 13:50
huf so i suppose there's no possibility that other people have more insight than you
Woodi why it is smell ?
huf Woodi: because list context. it's almost surely not what you want.
Woodi huf: pls enlight me :)
masak Woodi: what type would 'undef' have in Perl 6? 13:51
Woodi example:
masak Woodi: remember that it can be returned from any function, even those that return Str or Int or DateTime.
Woodi sub a{ return undef }
masak auugh 13:52
PerlJam masak: careful ... he'll just invoke "magic" with a handwave.
masak Woodi: you're not listening.
huf Woodi: sub x { return undef } if (x()) { works as expected } my @x = x(); if (@x) { surprise }
masak Woodi: why are you using 'return undef' there, and not just 'return'?
huf Woodi: but that's just explicit return undef, not really the core issue here
Woodi then it works for: scalars, hashes, lists, bools - it Perl philosophy
mux lol
people have just shown you how it doesn't really work for lists 13:53
huf Woodi: p6 has more types than that...
masak Woodi: are you not reading what people are telling you?
mux I love people who have the nerve to dispute what perl's philosophy is, with people who are the main contributors to the language
masak Woodi: a conversation has to be a bit of give-and-take. you have to react to what we tell you, too.
Woodi mux: then it should be better in v6 :)
masak hugs Woodi 13:54
Woodi thanx :)
masak I think I needed that more than you did.
PerlJam #perl6 ... where you get a hug even if we suspect you're a troll :)
mux group hug?
moritz [Coke]: oh, my fault. eval() in rakudo catches exceptions, in niecza it doesn't
Woodi mux: contribution does not matter, using Perl matters 13:55
masak Woodi: you're using it wrong :)
mux whatever
moritz [Coke]: and I suppose niecza is correcter here 13:56
Woodi PerlJam: you think i am trolling here ?
masak Woodi: *I* think you're trolling here. 13:57
Woodi: but I also think you are not aware you are.
PerlJam Woodi: in any case, if your subroutine is likely to be used in both scalar and list contexts, you should use just "return;" to mean "give nothing back", otherwise if you use "return undef;" in list context, you end up with a one element list containing a single undef value.
mux it's very hard to differentiate an ignorant person and a successful troll
Woodi i think i just have other thinking
mux I for one just think you have no clue what you're talking about
snarkyboojum I've learned something from this convo anyway, undef isn't typed enough for Perl 6, that seems to be a big part of it, so you return Mu or Nil or an undefined type object
masak what mux said.
I'm glad I reached snarkyboojum++ :) 13:58
snarkyboojum maybe not 100% right, but seems to be part of it :D
masak then this conversation wasn't a total waste.
snarkyboojum masak++ for general learnedness ;)
moritz niecza: try eval('1 1')
p6eval niecza v10-38-g23f1b2e: ( no output )
moritz niecza: eval('1 1')
p6eval niecza v10-38-g23f1b2e: OUTPUT«␤Unhandled Exception: Two terms in a row at (eval) line 1:␤------> 1 ⏏1␤␤ at /home/p6eval/niecza/lib/CORE.setting line 1938 (CORE eval @ 4) ␤ at /tmp/WlCFX9p4gO line 1 (MAIN mainline @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line…
Woodi no. return Any, or return Mu is just to many headache for language like Perl
moritz so, return; then 13:59
Woodi but i will leave this subject now
snarkyboojum Woodi: well just a return from a sub will return Nil anyway, so you're all good :)
masak Woodi: JUST USE 'return;' IN EITHER PERL 5 OR 6, YOU F-ING MORON
flussence blinks
PerlJam masak: Hmm. Do you need another hug?
masak Woodi: WE'VE BEEN TELLING YOU THIS FOR HALF AN HOUR, DAMMIT
huf ... how is returning anything a headache?!
wat.
masak PerlJam: hug didn't help. letting of steam might've. 14:00
PerlJam hugs masak anyway
masak thanks. :/
flussence well someone was going to do it eventually. 14:01
snarkyboojum I think Woodi is just grappling with the idea - no hard feelings eh :D
huf wish we had mst here ;)
masak I agree that might not have been very #perl6-y. not sure it helped either me or Woodi, either.
huf masak: it makes this channel more believable 14:02
masak Woodi: for what it's worth, I don't think you're a moron. I think in this particular discussion, you have comprehension problems.
huf you're all too friendly and patient
masak Woodi: as in, it's not us (or Perl 6), it's you.
moritz huf: it works on the vast majority of subjects in our secrect psychology study :-)
[Coke]: I'll try to fix some of that fallout tonight 14:03
masak I confess that shouting was my second recourse, after realizing that I don't have op powers.
moritz you... don't?
14:04 ChanServ sets mode: +o moritz, moritz sets mode: +o masak, moritz sets mode: -o moritz
masak thanks. 14:04
flussence kicking people is never the correct answer, unless you're making a ruby joke.
PerlJam I think Woodi is just victim of the same disease of those people who say "Perl 6 isn't Perl because it uses a . instead of a ->" (i.e., not a deep enough understanding of the essence of the language)
masak PerlJam: aye, probably.
moritz PerlJam: one problem is that Woodi is voicing strong opinions, but doesn't know much about perl 6, and has even admitted that here 14:05
masak PerlJam: it's very naive to come in, pick a feature, and start yelling "you promised you wouldn't change the good parts", though.
moritz PerlJam: it's a form of agression
snarkyboojum except in this case the issues of definedness and types is a little deeper than syntactic changes from -> to .
masak they are.
mux I think what's really naive (and actually moronic) is not being able to realize you don't have the necessary knowledge of something to actually argue with people who obviously know what they're talking about 14:06
snarkyboojum maybe I'm too charitable, but I reckon he was genuinely trying to grok it :D
masak fwiw, I still feel a little strange when I do 'return Str' or similar.
my bet is that not many people have run into this situation yet.
Woodi masak: this wasn't only about 'return'. i was about to eagerly dropping things that works ok in v5. perl5 ppls want perl6 too. and new ppls in Perl community, like you, do not understand that
moritz masak: yep, but it doesn't happen often
PerlJam Hmm.
moritz Woodi: fwiw undef wasn't eagerly dropped. It was dropped after years of trying to figure out what 'undef' actually means, and realizing it doesn't have a meaning 14:07
in Perl 6, that is
snarkyboojum Woodi: I guess the message was that there are good reasons for why the change has happened, and they're probably a little subtle
masak Woodi: I agree there's a general risk of changing "too much". but "undef" isn't such a case.
moritz masak: in the end you usually want to fail() instead of return Str;
masak aye.
I think so too.
PerlJam Woodi: everything that has changed in Perl 6 from Perl 5, has done so after quite a fair bit of thought. Sometimes they've changed several times before we've gotten where we are today.
masak Woodi: you've undermined yourself somewhat in this discussion by defending 'return undef' when people have been telling you (repeatedly) that it's probably an error to use it. 14:08
flussence two errors, actually: perl -wle 'sub x { return undef }; my %x = x()' 14:09
:)
masak Woodi: but even aside from that, 'undef' was changed for really good reasons, IMO.
Woodi moritz: first notice i was asking about how to do 'return undef'. before i tested 'return' probably someone advised to use 'return Any' and that is not what i expect form Perl. and then discussion started.
[Coke] wonders if we can get a perl5 mode for evalbot.
moritz [Coke]: buubot_backup has a perl 5 mode
buubot_backup: eval: undef + 2
buubot_backup moritz: ERROR: Can't modify constant item in undef operator at (eval 20) line 2, at EOF
PerlJam Woodi: in any case ... we're not the ones you need to convince if you want to make a change in Perl 6. We're not the language designer.
moritz PerlJam: we are :-) 14:10
at least a subset of them
PerlJam moritz: shh :)
masak PerlJam: didn't you get the memo? :P
moritz to /return/ to a more technical issue, I'm now trying to make rakudo not catch exceptions in eval()
it's been spec not to catch for some time
and niecza doesn't catch
masak ++morotz
augh
++moritz 14:11
moritz and the mismatch between the two causes maintenance nightmares in roast
snarkyboojum Niecza does perl5 bits and bobs now eh.. not sure if that's p6eval'able
moritz snarkyboojum: I'd rather not, it probably doesn't have a safe mode :-) 14:12
masak Woodi: we have types in Perl 6. that's the main reason 'undef' had to change.
snarkyboojum Woodi: lots of juicy reading in the perl6-language archives fwiw e.g. osdir.com/ml/perl6-language/2009-11...00075.html
dalek kudo/eval-throws: 741a933 | moritz++ | src/core/control.pm:
Do not catch exceptions from eval()
masak Woodi: that's also the main reason 'wantarray' had to go.
snarkyboojum moritz: good point :)
flussence while we're on an error-handling topic, I want to whine about rakudo's Test.pm: gist.github.com/1179681 14:13
PerlJam masak: They were warts on the flesh of context-sensitivity?
masak PerlJam: :) 14:14
moritz flussence++
snarkyboojum Woodi: or rather the real spec on definedness perlcabal.org/syn/S02.html#Undefined_types # I will stop now :D
masak PerlJam: there wasn't any way to combine MMD and wantarray. and we liked MMD more :) 14:15
dalek kudo/eval-throws: fa2d5c5 | moritz++ | lib/Test.pm:
Test.pm needs to deal with eval() not catching exceptions
 14:23
Woodi masak: types are additional feature to 'automagic typing' - where it is possible 14:24
moritz Woodi: that's bullshit
Woodi: types are very deeply rooted in the language
even if it doesn't always appear to be the case 14:25
Woodi moritz: i am just user
mberends Woodi: you might be interested in the history of the demise of undef. Here is the moment it became doomed: irclog.perlgeek.de/perl6/2009-11-24#i_1762914. Prior to that, it caused problems, for example irclog.perlgeek.de/perl6/2009-11-23#i_1760736. Mu and Nil were still to me named, as you will see if you read further.
Woodi and it works that way
moritz Woodi: if your argument is "i am just a user", don't participate in language design discussion
mux spot on.
flussence it helps if you've actually *used* the language too.
moritz Woodi: if you want to participate in those discussions, you can't be "just a user", but you need to actually understand the implications of stuff 14:26
masak that's what makes language design both thrilling and frustrating :)
usually the implications aren't understood until there's an implementation on which to try a feature out.
mux designing a language is science; not packing N 'cool' features together... 14:27
moritz feedback is ok, but then please formulate it constructively
PerlJam mux: I disagree. Designing a languae is art :-)
mux PerlJam: heh. and quite a bit of that too yeah ;-)
Woodi moritz: 1. i do not say anything autoritary. OK ? 14:28
mux that sentence alone just made my irony-meter explode
moritz Woodi: you basically say "$thing sucks". That's not constructive.
Woodi 2. real programmer feedback on Perl: types are additional feature to 'automagic typing' - where it is possible
snarkyboojum Woodi: as a user of a language type info can be very helpful.. perl 6 uses the idea of gradual typing - see 6guts.wordpress.com/2010/09/20/grad...alization/ 14:29
Woodi moritz: i was wrong. just 'return' is better :)
moritz Woodi: ok, great 14:30
Woodi snarkyboojum: it is what i say in details
snarkyboojum Woodi: there's lots of info out there for the inquisitive mind :)
moritz [Coke]: if you ack 'nok\s+eval' in roast, you get a list of moritzFAIL :-)
Woodi was doing tee when you still yelling on me :) 14:31
Woodi thinks forcing 'return Anything' is bad. but seems no one force this :) 14:32
PerlJam Woodi: Perl doesn't *force* anything. IT strongly encourages things, however. 14:34
Woodi: this is part of the essence of Perl. TMTOWTDI after all
benabik is in the middle of backlogging, but... 14:36
Does fail() return, or do you need to return fail?
moritz benabik: fail() returns
benabik wonders why he asks when he has perl6 installed. 14:37
masak benabik: that habit is hard to break. :)
at least you notice it, unlike most of p6l. :P 14:38
benabik Hm. The REPL isn't as robust as it could be. `sub foo { fail('bar') }; my $x = foo` just exited the REPL. It printed bar, but nothing else. 14:39
masak actually, p6l has been mostly silent recently. not really living up to its reputation as a think tank.
masak needs to re-ignite the 'rw attr' thread
moritz masak: it usually is quite the think tank if you inject a quesiton or idea
felher masak++ for macro-branch :D
masak benabik: we've had a bunch of those things. this particular one might be new. 14:40
benabik: want to report it to RT?
benabik masak: where's RT?
masak benabik: just email [email@hidden.address] about it.
benabik masak++ 14:41
masak like magic, it'll turn into a bug report in the system :)
14:43 DarkWolf84 joined
DarkWolf84 hello #perl6 14:43
masak DarkWolf84! \o/
DarkWolf84 hi, masak 14:44
masak felher: ooh, you've tried it out? any nice use-case yet? :)
benabik golfs a bit.
DarkWolf84 nom: my $a = sub { return }; say $a; 14:45
p6eval nom ff78b5: OUTPUT«_block1011␤»
DarkWolf84 is that corect?
I think it shoud be Any
benabik nom: my $a = sub { return }; say $a() 14:46
p6eval nom ff78b5: OUTPUT«Nil␤»
benabik DarkWolf84: The _block1011 is the (somewhat LTA) string for the sub.
PerlJam DarkWolf84: It should not be Any, but it certainly should not be "_block1011"
14:47 daxim left
benabik I think it's probably supposed to be Code. 14:48
snarkyboojum niecza is arguably more meaningful
niecza: my $a = sub { return }; say $a;
p6eval niecza v10-38-g23f1b2e: OUTPUT«{ ... }␤»
benabik Er, Sub.
snarkyboojum or not :)
benabik The parrot object is leaking through. 14:49
snarkyboojum I like it tho ;)
benabik nom: fail; say 'foo' 14:50
p6eval nom ff78b5: OUTPUT«foo␤»
benabik nom: sub foo { fail; say 'foo' }; foo
p6eval nom ff78b5: ( no output )
felher masak: Nope, sorry, had to learn for an exam. But i sure will. :) Will/did you blogpost about your branch?
benabik That seems odd.
masak felher: yes, as soon as I manage to get back on the horse again with blogging. 14:51
14:51 jaldhar left
masak somehow there's not as much time in a day as there used to be. 14:51
DarkWolf84 masak: ok, it was the braces 14:52
masak DarkWolf84: right, I should've told you. sorry :) 14:53
DarkWolf84 so perl6 is context language like ever :)
benabik More concise...
masak I looked at that and thought "he probably meant to call it". forgot to mention it, though.
DarkWolf84 s/like ever/like perl 5/
benabik nom: sub foo { fail; say 'foo' }; foo; fail; say 'bar' # I expect either "foo\nbar" or nothing, but get...
p6eval nom ff78b5: OUTPUT«bar␤»
masak DarkWolf84: it's not so much context here, as explicitly calling a variable. 14:54
DarkWolf84: to call &foo, you either have to do '&foo()' or 'foo'.
DarkWolf84: with '$a', there's only the '$a()' option.
DarkWolf84 masak: but foo is just a sub not a closure
14:55 kmwallio joined
masak DarkWolf84: the difference between sub and closure doesn't matter from the perspective of calling it. 14:55
benabik nom: my &foo = sub { return }; my $bar; say &foo; say $bar; say &foo(); say $bar(); say foo 14:56
p6eval nom ff78b5: OUTPUT«_block1011␤Any()␤Nil␤Method 'postcircumfix:<( )>' not found for invocant of class 'Any'␤ in <anon> at src/gen/Metamodel.pm:3164␤ in block <anon> at /tmp/JpNnxJP95D:1␤ in <anon> at /tmp/JpNnxJP95D:1␤»
benabik Oh, forgot a step...
nom: my &foo = sub { return }; my $bar = &foo; say &foo; say $bar; say &foo(); say $bar(); say foo
p6eval nom ff78b5: OUTPUT«_block1011␤_block1011␤Nil␤Nil␤Nil␤»
felher masak: Sounds good to me. :) 14:57
masak benabik++
felher: general plan: an introductory post about macros and what they are; then a post about the work with D1. 14:58
benabik _block1011 is still somewhat LTA. Code should probably .gist to '{ … }' or something
14:58 GlitchMr joined
benabik Not sure how it's falling back to the parrot sub. 14:58
PerlJam benabik: or ... the actual code :)
jnthn Because we want named subs to stringify to the name of the sub. 14:59
phenny jnthn: 06:17Z <japhb> ask jnthn Can you make '$value but True' and '$value but False' work without too much trouble? This is coming up when trying to implement MAIN as per spec.
jnthn: 06:18Z <japhb> tell jnthn sorear++ has found a workaround. A bit uglier, but the long form seems to work.
jnthn And it doesn't make sense to have our own slot for name when a Parrot sub has such a slot
DarkWolf84 nom: my $array_reff = <a b c>; $a.say; a.perl.say
p6eval nom ff78b5: OUTPUT«===SORRY!===␤Symbol '$a' not predeclared in <anonymous> (/tmp/EOPfxRFz_7:1)␤»
DarkWolf84 oops 15:00
jnthn Probably should be more like <anon> or so though, rather than _block1234
PerlJam jnthn: indeed.
DarkWolf84 nom: my $array_reff = <a b c>; $array_reff.say; array_reff.perl.say
p6eval nom ff78b5: OUTPUT«a b c␤Could not find sub &array_reff␤ in block <anon> at /tmp/tx20zgbwCP:1␤ in <anon> at /tmp/tx20zgbwCP:1␤»
jnthn I can probably do something about it; file a ticket.
PerlJam DarkWolf84: missing a sigil
jnthn decommute &
DarkWolf84 again 15:01
nom: my $array_reff = <a b c>; $array_reff.say; $array_reff.perl.say
p6eval nom ff78b5: OUTPUT«a b c␤("a", "b", "c")␤»
DarkWolf84 ok
felher masak: Nice. *waitingForIt* :)
15:01 SHODAN left
DarkWolf84 I really must try nom :) 15:02
TimToady why do we want to stringify to just the name, when .name is available for that?
15:03 baest left
DarkWolf84 PerlJam, I'm verry absend theese days 15:04
15:06 risou_awy is now known as risou
DarkWolf84 absent* 15:06
TimToady in fact, I would argue that sub foo might stringify to &foo, and an anonymous function to just &
masak cute.
perl6: sub foo {}; say &foo ~~ &foo 15:07
p6eval rakudo ff78b5, niecza v10-38-g23f1b2e: OUTPUT«Nil␤»
..pugs: OUTPUT«␤»
masak hm.
benabik Nil?
DarkWolf84 O_O
masak perl6: sub foo { say "because it was run" }; say &foo ~~ &foo
p6eval rakudo ff78b5, niecza v10-38-g23f1b2e: OUTPUT«because it was run␤Bool::True␤»
..pugs: OUTPUT«because it was run␤1␤»
DarkWolf84 yeah i got it 15:09
nom: say Any ~~ Any 15:10
p6eval nom ff78b5: OUTPUT«Bool::True␤»
DarkWolf84 nope
nom: say 0 ~~ 0 15:11
p6eval nom ff78b5: OUTPUT«Bool::True␤»
DarkWolf84 ...
oh, the refs are different 15:12
TimToady smartmatching is not symmetrical
moritz and nearly everything is Any 15:13
DarkWolf84 i forgot ~~ is not ==
masak nom: say Junction ~~ Any
p6eval nom ff78b5: OUTPUT«Bool::False␤»
masak .oO( "I said *nearly* everything!" ) 15:14
TimToady nom: say Any ~~ :!defined
p6eval nom ff78b5: OUTPUT«Bool::True␤»
DarkWolf84 why Any is defined
moritz it is not
TimToady it's not
moritz that's the !
TimToady or rather, it's !
benabik nom: sub foo { 2 }; say 2 ~~ &foo 15:15
p6eval nom ff78b5: OUTPUT«2␤»
benabik nom: sub foo { 2 }; say 1 ~~ &foo
p6eval nom ff78b5: OUTPUT«2␤»
DarkWolf84 my dad
masak are you talking about !?
DarkWolf84 my bad*
moritz nom: say so try { eval 'die "foo"' }
p6eval nom ff78b5: OUTPUT«Bool::False␤»
moritz I get a segfault with my local patch :( 15:16
nom: try { die "foo" }; say so $!
p6eval nom ff78b5: OUTPUT«Bool::True␤»
moritz am I right in assuming that exceptions should be False rather than True? 15:17
TimToady benabik: you're exercising S03:3611
benabik TimToady: I would almost expect a 0 arg sub to be called and then the result smart matched against... 15:19
Ignoring the LHS of ~~ is odd. (to me)
TimToady the left arg was already bound to $_, and the right arg was *already* called for the smartmatch, and the smartmatch found a Callable on the right, so it's entirely up to the Callable whether it pays attention to $_ or not 15:21
masak moritz: I can't think of a reason they should. but maybe there is one?
benabik TimToady: Fair enough. 15:22
masak moritz: also -- False and :!defined, or False but :defined?
moritz masak: not sure about defined 15:25
DarkWolf84 will qregex implement LTM 15:26
?
moritz masak: my use case is that I want to write nok try { eval $string }
masak: and try { } should return the exception on failure, afaict
or should it return a Failure that wraps the exception?
and then the Failure is neither .defined nor True?
tadzik wow, there have been shouts
moritz tadzik: not only shouts, but shouts by masak :-) 15:27
tadzik I can barely imagine masak shoulting
or even shouting
DarkWolf84 me too
masak me neither.
tadzik hugs masak 15:28
DarkWolf84 the trols make mirakles sometimes
masak ...but if it gets me this many hugs, maybe I should start shouting more often... :)
tadzik I found it more like misunderstanding than trolling
masak oh, it was.
tadzik SOMEBODY HUG ME DAMNIT
:)
masak a misunderstanding, I mean. 15:29
suddenly I found myself thinking of mst's post about "sometimes the only way to reach through to a person that wastes everybody's time is to shout at them".
and then I tried it.
benabik masak: I'm not sure WWmstD is a good way to live your life. ;-) 15:30
masak here's the post, by the way: www.shadowcat.co.uk/blog/matt-s-tro...a-bastard/
benabik: heh :)
tadzik lecturestime, will be a bit early and a bit late for #phasers today
benabik obey, I don't have a lecture today. I can be in #phasers! 15:32
15:33 tzhs left
DarkWolf84 ok, bb then 15:40
15:41 DarkWolf84 left 15:42 mkramer left 15:43 mkramer joined, molaf left
im2ee Catalyst, Dancer or Mojolicious? :) 15:47
Which is the best one?
flussence I'd go with mojolicious, but I haven't used the middle one 15:48
masak if I were to start a project today, I'd probably reach for Dancer.
but I have nothing against the other two. 15:49
im2ee masak, so why Dancer? :)
flussence catalyst is like using a mountain to swat a fly
im2ee I'm looking for the easiest one.. I won't spend a lot of time for web development. :)
masak im2ee: Dancer seems very easy to get started with. 15:50
flussence try.rakudo.org's frontend uses mojolicious, but it doesn't really do much with it besides render one page.
masak Catalyst seems to be very capable and have a dedicated community. the Catalyst community is likely the biggest of the three. 15:51
im2ee So, i probably will use Dancer. :) 15:52
flussence use all of them at the same time!
im2ee flussence, maybe i will. I must think it over. :) 15:53
flussence
.oO( hm, nobody's written an Acme::Cancerlicious yet )
 15:54
15:59 packetknife joined 16:00 colomon joined
masak decommutes 16:01
felher If i have a array of hashes like { value => X, name => Y }, is there a common idom to get the maximum hash compared by hash<value>? 16:03
moritz felher: max :by({ $^a<value> <=> $^b<value> }), @hashes 16:04
[Coke] oooh. my coworker bought me some spicy padam. mmmm. 16:05
16:13 MayDaniel joined
PerlJam Would a Perl 6 version of Dancer be called Prancer? (and would that give the wrong impression? ;) 16:13
im2ee PerlJam, why Prancer? :) 16:14
PerlJam why not? 16:15
im2ee ;)
PerlJam im2ee: btw, since you were polling, Dancer++ :)
im2ee So i will test it. :) 16:16
karma Dancer
aloha Dancer has karma of 1.
moritz im2ee: Mojolicious++
im2ee moritz, why? :)
karma Mojolicious
aloha Mojolicious has karma of 1.
im2ee :)
The same karma. :)
moritz im2ee: because it works, and is fun 16:17
im2ee: and much easier to get started with than Catalyst
(ok, that's not too hard :-)
PerlJam I've never used Mojolicious but from what I've read, it seems kind of like Catalyst re-imagined to be more like Dancer.
im2ee moritz, have You ever been using Dancer? :)
_sri mojolicious is more for people that love web development, not so much for those that just want to get away from it again as quick as possible (imo) 16:18
a key feature is real-time web apps
PerlJam _sri: yeah, it seems to have that "fun" thing going for it :)
moritz im2ee: no
im2ee: which is why I don't comment on it 16:19
PerlJam is hacking a Dancer app right now.
_sri i personally consider mojolicious my catalyst2 16:20
im2ee moritz, so.. i'll try both. :)
felher moritz: Thnx, but thats what i tried first (well, i actually tried @hashes.max: { $^a<value> cmp $^b<value } ), but it seems that one of the first two values that getting compared is -Inf, and Num<value> doesn't work :)
moritz felher: then try .sort({block}).[*-1] 16:21
_sri but yea, you should love web development, or you won't appreciate either 16:22
felher moritz: k, thnx :) My array is small enough so sort doesn't cost much, i guess :) 16:23
16:23 wolfman2000 joined
colomon #Phasers in 67? 16:24
PerlJam _sri: what's a "real time web app" ?
16:25 daniel-s left
_sri PerlJam: en.wikipedia.org/wiki/Real-time_web 16:25
Comet/WebSockets 16:26
PerlJam _sri: so, how does mojo support "real-time web" exactly? Just that it's fast and unobtrusive? 16:27
slavik1 real time web = turning web browsers into thick clients ...
moritz PerlJam: websocket support
[Coke] (dancer,mojo,catalyst) I picked up mojo because of the screencasts, and it has been very very nice to me so far. 16:28
mberends 2
PerlJam [Coke]: I started watching them too, but I haven't done anything with Mojo yet.
16:28 mkramer left
[Coke] PerlJam: if you wish to play, see: github.com/coke/muddle and maybe commit some stuff. ;) 16:29
ugh, haven't touched it since 9/21.
16:31 envi_ left
_sri PerlJam: supporting Comet and WebSockets properly is really hard, our whole stack is non-blocking/evented 16:32
PerlJam _sri: gotcha. (now that I've read a little more :)
16:33 envi_laptop joined
felher nom: my @array = 1,2,3; @array = @array 16:38
problem known?
p6eval nom ff78b5: OUTPUT«maximum recursion depth exceeded␤ in method infinite at src/gen/CORE.setting:4012␤ in method infinite at src/gen/CORE.setting:3710␤ in method infinite at src/gen/CORE.setting:4022␤ in method infinite at src/gen/CORE.setting:3710␤ in method infinite at src/gen/COR…
16:40 alester left
[Coke] felher: "doctor, it hurts when I do this!" 16:41
but yah, that's a bug.
felher [Coke]: Sorry, i'm not sure i understand what you wanted to say with "doc..." :) 16:44
16:46 daniel-s joined
moritz submits rakudobug 16:49
[Coke] felher: man goes to the doctor, having hurt his arm in an accident. He shows the doctor, moving his arm, saying "doctor, it hurts when I do this!" the doctor says, "well, don't do that! next!" 16:51
felher [Coke]: Ah, i see. Thnx for explaining :) 16:52
[Coke] overtells it. I think the original is just: 16:53
The patient says, "Doctor, it hurts when I do this." "Then don't do that!"
(henny youngman)
huf it's explained/documented in the jargon file :D 16:54
[Coke] finds a list of henny youngman jokes on the internet and is distracted. 16:55
16:55 dakkar left
[Coke] A man goes to a psychiatrist. "Nobody listens to me!" The doctor says, "Next!" 16:56
huf "doctor, it hurts when i pee" "sir, there's a knot on your wiener" "oh damn, i forgot the bread!" 16:57
benabik Bread?
huf tying knots on your hankie is a way to remember to do things
according to anecdotes
i'll keep telling this joke until the world learns about this custom 16:58
:D
benabik ah
moritz problem is, it doesn't tell you *what+ you have forgotten :-) 16:59
huf moritz: well, the guy in the joke figured it out eventually :D
jokes translate rather badly between even slightly different cultures :( 17:00
sjohnson does Perl6 do a lot of the fancy IPC stuff? 17:01
flussence I think the fanciest stuff we have is qx()... 17:02
17:03 donri left, nebuchadnezzar left
moritz jnthn: regarding the optimizer, a shorthand for nqp::want(nqp::p6box_i(some_expression_returning_int), 'Ii', same_expression_returning_int) would really be nice 17:10
same for the Num/num case
masak huf: similarly, it's more effective to remember how many pieces of luggage you're carrying on your travels, rather than which ones.
TimToady With respect to genericity/specificity, most programmers have been taught either to be noun-dominant (OO) or verb-dominant (FP), but Perl 6 tries to keep both of those in different parts of its brain, just like you do. 17:11
huf masak: i was told that's how to remember your shopping list 17:12
jnthn moritz: Yeah, I did what I could with the things I had to hand. I don't find it ideal, I just don't know exactly what better looks like yet. :)
masak huf: yeah. I do that too. sometimes I make mnemonics out of the shopping items, though.
TimToady Much of the pushback on the design of Perl 6 has come from people who don't want to rethink their verbs vs nouns.
see undef vs undef, for instance
huf masak: i tend to list in a sea of fog, as far as memory is concerned
TimToady of wantarray (verb) vs Parcel (noun)
*or 17:13
moritz jnthn: nqp::p6autobox_i(expression_returning_int) maybe?
17:13 MayDaniel left
TimToady or == as defined by nouns rather than by the language :) 17:13
masak im2ee: thanks for starting the nice discussion on web frameworks. :)
jnthn moritz: Yeah, I was trying to find a way that fits in with what map_node already does
moritz: But maybe that's a less important goal.
masak huf: huh! my mental space isn't foggy, it's a whiteboard. which is funny, since I prefer blackboards in meatspace. :) 17:14
moritz jnthn: it could compilre rather directly to what we have now
jnthn moritz: Yes, I could make it do so :)
masak huf: but I usually make lists on my fingers, right hand first.
huf masak: in kindergarten, i was convinced (for years) that each day was two because we had an afternoon nap in the middle 17:15
not much has changed as far as my brain is concerned
moritz jnthn: on a related matter, how much sense does it make to add native ops that return Bool?
jnthn: ie infix:<==> and the like
huf masak: if it's not in my irclogs, it's mythical :D
diakopter backlogs to see why masak is @
pmichaud good afternoon, #perl6
flussence it's a halo
PerlJam diakopter: he was feeling very #perl-like earlier
jnthn pmichaud! \o/
17:16 am0c joined
moritz good pm, pm 17:16
17:17 Chillance joined
masak pmichaud! \o/ 17:17
diakopter: I lost my temper, so they gave me op. 17:18
17:18 bluescreen10 left, bluescreen10 joined
[Coke] ho, pmichaud++ 17:19
pmichaud nom: my &foo = /bcd/; say 'abcdefg' ~~ /<&foo>/;
p6eval nom ff78b5: OUTPUT«=> <bcd>␤␤»
moritz nom: say so 'ab' ~~ /^.*b/ 17:20
p6eval nom ff78b5: OUTPUT«Bool::True␤»
moritz nom: say so 'ab' ~~ /^(.*)b/
p6eval nom ff78b5: OUTPUT«Bool::False␤»
moritz that one
17:22 rml left 17:24 rml joined
diakopter wow, several people got very huffy/defensive 17:25
17:25 dark_x joined
masak diakopter: "defensive" is the word you generally use when these discussions happen. it's not a word I would use. what do you mean by it? 17:27
im2ee masak, all my pleasure. :) 17:29
afk
moritz #phasers time 17:30
TimToady diakopter: I saw little huffy/defensive, but a great deal of frustration with the inability to communicate effectively
PerlJam Yeah, more "annoyed" and "exasperated" and "irked" 17:31
GlitchMr nom: print
0􏿽xB2
p6eval nom ff78b5: ( no output )
GlitchMr nom: print 0􏿽xB2
p6eval nom ff78b5: OUTPUT«===SORRY!===␤Confused at line 1, near "print 0\ufffd"␤»
GlitchMr Accidental line break character
\uf 17:32
\uffd
Yeah, it makes sense
I have wrongly configured IRC client
nom: print 0􏿽xB2
p6eval nom ff78b5: OUTPUT«===SORRY!===␤Confused at line 1, near "print 0\ufffd"␤»
17:33 GlitchMr left
diakopter TimToady: ok 17:33
TimToady myself, the discussion both here and and on #parrot led me to muse about nouns and verbs, and how hardwired they get in our brains. :) 17:35
masak ooh, there was a discussion on #parrot?
pmichaud ...I missed a discussion? There was one that I didn't instigate/throw gasoline upon? ;-) 17:36
TimToady about the meaning of ==
it was just a small conflagration :)
masak :P
diakopter masak: I guess I meant that I saw people taking [their understandings of] Woodi's questions/statements personally, as if they thought Woodi was trying to make assertions/claims that he doesn't deserve to make.. so I interpreted folks' responses as trying to defend their views that Woodi has no right to make such assertions/claims
17:36 bluescreen10 left
masak diakopter: ooh, that *is* a good use of the word. 17:37
diakopter: as for me, you're right in the sense that I let it get to me.
TimToady otoh, observing that the path to understanding proceed through the step of understanding that you don't understand is not entirely a defensive notion :) 17:38
*ceeds
masak diakopter: but it wasn't so much "doesn't deserve to make" as "ok, how do I get this guy's attention?"
PerlJam masak: and then you turned into mst for a brief moment :) 17:39
masak next time, if there's a next time, I won't call the person a moron, but the act.
having read mst's post more closely, that seems to be what he does.
which makes a lot of sense.
17:39 ashleydev joined 17:40 nebuchadnezzar joined
masak PerlJam: well it was half experiment and half venting of frustration. 17:40
PerlJam yes, and ad hominem statements tend to escalate any ... misunderstandings rather than quell the situation.
masak nod.
17:40 bluescreen10 joined
masak to Woodi++'s credit, he took it well (partly because he was away making tea). 17:40
TimToady it did appear that the person in question does (did?) not yet understand the extent to which he does not understand... 17:41
but we all have areas where we're like that... 17:42
masak TimToady: not sure I understand... :) 17:43
diakopter masak: did you say something?
masak .oO( Meta-Dunning-Kruger )
17:49 MayDaniel joined
masak though Dunning-Kruger is pretty meta already. 17:49
dalek kudo/optimizer: ad8f64a | moritz++ | src/core/Int.pm:
more native int ops
masak at least as far as I understand it :)
17:56 bluescreen10 left 17:59 daniel-s left 18:05 MayDaniel left
Woodi i think i was scared about USABILITY eg. forcing ppls to choose what undefined type they should return; but probably others answered to perl5-undef-implementation versus perl6-undefs-idea-refactoring... i must agree i should not be allowed to talk about second... 18:06
but about versus vs nount - perl5 undef is (probably) both... 18:07
TimToady we might not have chosen to detangle the verb from the noun, except that Perl 6 has to be much pickier about whether the next thing is a term or not, and we aren't allowed to guess about that, as Perl 5 does 18:10
18:11 bluescreen10 joined
TimToady in order to be extensible, Perl 6 relies much more heavily on catching two-terms-in-a-row errors 18:12
and ambiguous verbs/nouns tend to defeat that
taking that together with the fact that there are many different ways of being undefined, there really wasn't much point in keeping undef 18:14
as for the panic, we've had a saying from the start of Perl 6, which is that "Everyone is allowed to panic, once." 18:15
:)
interestingly, masak++ is one of the chief violators of that rule, but we keep him around anyway for the entertainment value. :) 18:16
masak I was wondering about my panic count at this point. 18:17
tadzik he still has something twisted before on his head
Woodi TimToady: but i still have problem with 'return;' returning something undefined. what if before that $_ have some defined value ? 18:18
TimToady return as nothing to do with $_ in either Perl 5 or Perl 6, so I don't understand your question.
s/as/has/ 18:19
Woodi 'return undef' was 'scripting language trick'
i assume: if no walue then $_ is used. maybe i am wrong
PerlJam Woodi: you are (wrong). 18:20
TimToady you've made a false generalization about Perl 5
only certain functions default to $_
return is not one of them
Woodi ok
TimToady we decided that having to memorize a list was bad
so in Perl 6, *no* functions default to $_
you can use an invocant-less method if you like, though 18:21
Woodi then how [$@%]_ work ?
masak nom: sub defsay(*@a) { if @a { say @a } else { say $_ } }; $_ = 42; defsay
p6eval nom ff78b5: OUTPUT«Any()␤»
tadzik zing!
masak hm... $_ not dynamic...
PerlJam Woodi: what do you mean? 18:22
18:23 replore left
TimToady there are still some special syntactic constructs that default to $_ 18:23
such as, .foo means $_.foo
and 'when' implicitly smartmatches against $_
and s/// still works on $_ by default
but mostly, those variables are considered special self-declaring parameters now 18:24
masak rakudo: $_ = 42; .++; .say
p6eval rakudo ff78b5: OUTPUT«===SORRY!===␤Method 'unshift' not found for invocant of class 'Undef'␤»
masak hah!
masak submits rakudobug
b: $_ = 42; .++; .say
Woodi ok
p6eval b 1b7dd1: OUTPUT«===SORRY!===␤Method 'unshift' not found for invocant of class 'Undef'␤»
jnthn masak: already ticketed, I think :)
masak curses! I probably beat myself to it! 18:25
TimToady I think I asked about .++ last week, in fact
masak oh yeah
18:25 kaare_ joined
TimToady Woodi: so constructs that hardwire $_ in Perl 5 don't necessarily have to do that in Perl 6; for instance, when you say grep { $_ eq 'foo' } 18:26
in Perl 5, it's a special syntax that finds a block and forces $_ to be a kind of loop parameter
in Perl 6, it's just an anonymous function that is the first argument to grep 18:27
and when you write { $_ eq 'foo' } it's as if you'd said sub ($_) { $_ eq 'foo' }
Woodi so $_ is now near plain variable... 18:28
PerlJam Woodi: same with @_ and %_
Woodi ok, memoized i think 18:29
PerlJam Woodi: Perl 6 is simultaneously less magical and more magical than Perl 5. :) 18:30
18:30 envi_laptop left
Woodi PerlJam: oh yes, sometimes i think Perl6 is to compressed by Huffman method :) 18:31
TimToady that is one design principle out of many, but we do use it from time to time 18:32
masak any program or language can be ruined by taking a good thing to extremes.
PerlJam masak: look at lisp! :)
dalek kudo/optimizer: b6c5071 | moritz++ | src/core/Num.pm:
native num trig ops
kudo/optimizer: e19f726 | moritz++ | src/core/Num.pm:
more num builtins, including log() and the rest of the trig functions
masak PerlJam: hey, careful. do not upset the big-bearded ancestors! 18:33
Woodi TimToady: i worry that so many new features can scary a lot of users... but if someone dig that he/she drown :) 18:34
PerlJam Woodi: Similar to how you can program in Perl using metaphorical baby-talk, you can pick up the "new features" piece-meal so that they aren't so onerus. 18:36
TimToady keeping new users from drowning in too much detail is also one of our design principles
but it's all tradeoffs 18:37
there's no such thing as a perfect language
masak awww
TimToady and that's the fault of the waterbed theory, which is another design principle
PerlJam TimToady: context is king! Perfect for what purpose? :)
TimToady the waterbed theory is actually implicit in huffman coding, since you can only make some things easier by making other things harder 18:38
Woodi PerlJam: in Perl context the usual context is UNIX admins scripting (<6)... 18:39
tadzik I won't be so sure
TimToady another design principle in Perl 6 is that there are always *many* contexts, often with conflicting goals
PerlJam Woodi: maybe. There are lots of perl-based web apps out there. 18:40
Woodi: And I primarily use perl at work for data processing and number crunching stuff (with PDL)
masak Woodi: at $dayjob, it's web apps all the way. not so much UNIX admins as everything admins. :) 18:41
TimToady even for a single piece of software, the users, designers, programmers, marketers, and administrators can have profoundly conflicting goals.
Woodi TimToady: i have one dream... High Performance Perl... 18:42
TimToady Perl 6 is much likelier to get there than Perl 5 ever could be 18:43
flussence
.oO( never had speed issues running frozen-bubble on my pentium 2... )
TimToady but high performance depends on good optimizers, and good optimizers depend on knowing a lot about your program 18:44
masak you make it sound so easy ;)
flussence but if someone implements perl6 on top of LLVM or the like, then it becomes Someone Else's Problem :)
TimToady so an optimizable language makes many distinctions, which worries Certain People 18:45
flussence: as we have seen already, if the VM in question doesn't know what to do with the distinctions provided by the language, not much optimization can happen ;) 18:47
18:47 bluescreen10 left
masak that's the challenge of optimization: it cuts through all layers of the stack. 18:48
a few other things do that too. DESTROY submethods among them. we don't have those either.
PerlJam It would be interesting if Perl could do some type-inference and tell you when you've used an object in a slot with an incompatible type
jnthn Well, I see it more as being able to pass sufficient information down the stack.
TimToady and why the impedance mismatch with the parrot model was such a killer
jnthn PerlJam: It's on my todo list for optimizer.
TimToady anyway, all that <waves hands upward> is another design principle 18:50
we've got 'em coming and going
18:50 birdwindupbird joined
Woodi i would like that one day marketing ppls will go down to admins and say: we want you use Perl in next app becouse we do not want spend money on new servers but we want new projects perform better on old servers :) 18:50
18:50 birdwindupbird left
PerlJam Woodi: I don't think that marketing people care about new servers or not :) 18:52
Woodi PerlJam: they do... it is pain for them - how to buy as small amount as possible :) 18:53
18:55 skangas left 18:58 GlitchMr joined
Woodi ah, right, management ppls 18:58
19:00 bluescreen10 joined
im2ee re 19:00
moritz jnthn: fwiw, nom merged into optimizer runs mandelbrot-color in 37.8s
jnthn ooh, we broke 40s \o/
moritz: Is that unmodified mandelbrot-color? 19:01
moritz jnthn: yes
jnthn moritz: e.g. no extra type info added?
moritz: OK :)
nom: say (16 * 60 + 14) / 37.8
p6eval nom ff78b5: OUTPUT«25.7671957671958␤»
jnthn Something of a speedup. Still plenty of work to go :) 19:02
moritz 27.6s with additional type info
jnthn Wow. :)
Living the gradual typing dream :) 19:03
...well, not quite yet. :)
Woodi so if i talk so much today then two more sentence do not change to much i think :) so: Simple::Redis can store now keys, hashes and list; to be found in: github.com/slunski/perl6-simple-redis
moritz nom: sub f(int $x) { $x = $x + 1}; my int $x = 10; f($x); say $x
p6eval nom ff78b5: OUTPUT«10␤»
Woodi gratulation Perl6 magicans :) 19:04
moritz jnthn: there's probably no good way to make ++ work on native ints, is there?
Woodi and gn and gl #perl6
jnthn moritz: I suspect we have to make native int rw params work somehow.
moritz: It's non-trivial and I need to fathom how to do it in a way that optimizes down nicely. 19:05
moritz: So, ENOTYET.
But I really don't think telling people they can't ++ an int is going to fly.
moritz jnthn: any objections to me pushing the merging of nom into optimizer?
jnthn moritz: Go ahead. 19:06
TimToady C can ++ an int, so can we...someday...
moritz well, it could be macro-ish 19:07
dalek kudo/optimizer: 2146e92 | moritz++ | / (14 files):
Merge branch 'nom' into optimizer
jnthn Well, inlining practically is about turning a sub into a macro :) 19:09
TimToady it does imply understanding the nativeness of references though
jnthn TimToady: yes, that's what makes it a bit tricky :)
TimToady and how type information on a native variable is carried by its context 19:10
not by the variable itself
moritz jnthn: fwiw gist.github.com/1265348 doesn't seem to inline the call to &fancy, even though there's log(num) multi now
jnthn Well, that's what makes me not so scared of native references.
moritz jnthn: not sure if that's expected or not
jnthn moritz: It won't inline things with sub calls in just yet. 19:11
moritz: That needs some special care.
The optimizer todo list is....long :)
moritz and even weireder, the call is 19:12
$N100 = "&fancy"($P103)
I would have expected it pass a num register to &fancy
jnthn Should be possible too.
diakopter TimToady: since I've moved to California, I've notied people have stopped using linking verbs, definite article, and sentence subjects, and not just in IRC/IM/emails, speech also. Is this a California/SiValley thing or global or what?
moritz expects full awesomeness :-)
jnthn moritz: That's where I'm aiming :) 19:13
Rome wasn't built in a day :)
diakopter or 11 years 19:14
jnthn :P
ooh, it's stopped raining...I should visit the store and get something to eat for supper
moritz rome wasn't built in software either :-)
jnthn bbs
diakopter TimToady: is it just that any word that could possibly be left out unambiguously is left out, regardless of convention? 19:16
TimToady I would have to see some examples in situ, but I would hazard a guess that in such cases redundancy is provided out-of-band, or by protocol 19:18
you can't have effective human communication without some redundancy or error correction mechanism 19:19
so arguably, what we could be seeing is people not caring whether they communicate effectively :)
you could also be seeing spillover from people's native tongues 19:20
not many languages have definite articles, for instance
diakopter from $work irc: bruce: non symlink versioning live in production! # implied "is now"
benabik … symlink based versioning? 19:21
TimToady many languages omit "is" where it's obvious, such as classical greek
but I'd also argue that the ! has some aspectual, er, aspect to it 19:22
something like "anything surprising is by definition new information"
so that takes care of the "now"
diakopter PersonA: Existing project works for me. # implied "An" 19:23
TimToady indefinite articles are even rarer in human languages
diakopter PersonB: ok works on whatever server i'm on # implied "it" 19:24
it might be my company culture (to which I'm just not as accustomed as I might be) :/
TimToady there's a long tradition, largely unrecognized by grammar nazis, of leaving the subject out in English when it's obvious 19:25
"gotta run"
"sure is dark in here!"
"don't get no satisfaction" 19:26
colomon TimToady: Don't know what you are talking about
masak diakopter: I find those utterances terse but not surprisingly so.
colomon ;)
TimToady "works for me" 19:27
diakopter have to lunch & ;)
19:28 mberends left
masak articles are nice to have in a language but are by no means necessary. 19:28
their use is much more complicated than first meets the eye.
TimToady s/ a / 19:29
not to mention /
masak :) 19:31
19:31 MayDaniel joined
TimToady diakopter: note also the 1st/2nd person vs 3rd person distinction is partially carried by whether the verb has an 's' on it 19:36
"works for me" vs "work for me" 19:37
19:37 bluescreen100 joined
TimToady the latter can be either 1st or 2nd person depending on the question it's answering 19:37
"What do you do for a living?" "Work for me." 19:38
"What do you want me to do?" "Work for me."
19:38 MayDaniel left
TimToady when you say "work for me" to mean "work for myself" is also a fascinating linguistic topic. 19:39
why is it "I've gotta be me!" and not "I've gotta be myself!" ? 19:40
diakopter or "I've gotta be I"
TimToady fake latin doesn't impress me :) 19:41
diakopter I thought I was going to lunch
jnthn Curious. "Work for me" sounds really odd to my ears; I'd expect it to be "Work for myself"
TimToady didn't say you had to go
it's a mechanism for treating yourself as some other person, pragmatically speaking 19:42
linguistic multiple personalities 19:43
not just "me" 19:44
"Just look at you!"
why isn't it "youreself"?
s/e//
moritz in German we're much more picky about such reflexive forms
TimToady s/ about.*// :)
moritz :-)
though some of us are abolishing the Genetive, which is a realy pity 19:45
TimToady well, English is slowly losing the subjunctive, and "whom" 19:46
masak genitive* 19:47
TimToady the past is unevenly distributed too
masak from 'gignere', 'to beget'.
heh, sometimes Latin looks like ancient Italian :P 19:48
[Coke] wishes his phone would stop reminding him about the sixperl call. 19:52
masak those still happen? 19:53
19:54 MayDaniel joined
TimToady I would not infer that, were I you. 19:56
masak I see. 19:57
I now feel fairly confident in inferring that TimToady no longer participates in sixperl calls :P 19:58
PerlJam They could spontaneously start again ... :) 19:59
moritz calls "sixperl"
diakopter #phasers is more fun
moritz there you go, spontanesous sixperl call
masak groans
nom: say "spontaneous ", <one two three four five six>.roll, "perl call" 20:00
p6eval nom ff78b5: OUTPUT«spontaneous fourperl call␤»
TimToady Refrigerator!
[Coke] eek. 20:01
m6locks nice
[Coke] bananaphone call.
masak maybe I should've {}'d inward, rather than "-ing outward.
nom: say "ba", "na" x 2, "phone call" 20:02
p6eval nom ff78b5: OUTPUT«bananaphone call␤»
masak nom: say "ba{"na" x 2}phone call" 20:03
p6eval nom ff78b5: OUTPUT«bananaphone call␤»
masak finally embraces {} in qq strings as something he likes :)
TimToady++
sjohnson nom: say ("ring" X 7).join("-"), "... banana phone!" 20:04
p6eval nom ff78b5: OUTPUT«ring-7... banana phone!␤»
sjohnson nom: say ("ring" x 7).join("-"), "... banana phone!"
p6eval nom ff78b5: OUTPUT«ringringringringringringring... banana phone!␤»
masak sjohnson: 'x' is string rep, 'xx' is list rep.
sjohnson nom: say ("ring" xx 7).join("-"), "... banana phone!"
p6eval nom ff78b5: OUTPUT«ring-ring-ring-ring-ring-ring-ring... banana phone!␤»
sjohnson masak++
benabik nom: say "{join ('ring' xx 7): '-'}… ba{'na' x2}phone!" 20:05
p6eval nom ff78b5: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 1␤»
benabik nom: say "{('ring' xx 7).join('-')}… ba{'na' x2}phone!" 20:06
p6eval nom ff78b5: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 1␤»
benabik Ah well.
masak benabik: 'x 2', not 'x2'
STD has a better error message.
benabik nom: say "{join ('ring' xx 7): '-'}… ba{'na' x 2}phone!"
p6eval nom ff78b5: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 1␤»
masak aww :/
benabik std: say "{join ('ring' xx 7): '-'}… ba{'na' x 2}phone!" 20:07
p6eval std 20ae3bd: OUTPUT«ok 00:01 123m␤»
tadzik hehe
masak I'm not sure Rakudo supports indirect methods.
tadzik way better
sjohnson （　｀ー´）
benabik nom: say "{('ring' xx 7).join('-')}… ba{'na' x 2}phone!"
p6eval nom ff78b5: OUTPUT«ring-ring-ring-ring-ring-ring-ring… bananaphone!␤»
masak \o/
benabik masak++
masak b: say "{join ('ring' xx 7): '-'}… ba{'na' x 2}phone!"
p6eval b 1b7dd1: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 22␤»
20:07 kmwallio left
masak b didn't either. 20:07
moritz nom and b mostly parse the same 20:08
masak niecza: say join ('ring' xx 7): '-'
p6eval niecza v10-38-g23f1b2e: OUTPUT«===SORRY!===␤␤Invocant handling is NYI at /tmp/PTv0Ncv1UF line 1 (EOF):␤------> say join ('ring' xx 7): '-'⏏<EOL>␤␤␤Unhandled Exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 766 (CORE die @ …
masak awww.
moritz niecza++
masak but... who will implement it FIRST? :D
tadzik YAPSI!
20:08 y3llow left
masak well, Yapsi doesn't to methods yet... 20:08
20:09 pothos left
tadzik nom: my @a = "hip" xx 2; say ~(@a, @a.WHAT.perl), "!" 20:09
masak nor classes and objects in general...
p6eval nom ff78b5: OUTPUT«hip hip Array!␤»
masak lol
I had forgotten about that one ;)
tadzik you forget about it the second time, and for the second time I'm here to remind you that it's your invention :P 20:10
diakopter niecza: say: say 4; goto say; # TimToady why isn't say allowed as a label
20:10 y3llow joined
p6eval niecza v10-38-g23f1b2e: OUTPUT«===SORRY!===␤␤Unsupported use of bare 'say'; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument at /tmp/G9CQevIUQU line 1:␤------> say: say⏏ 4; goto say; # TimToady why isn't say a␤␤Two t… 20:10
masak tadzik: well, this time I (vaguely) remembered that I had originated it :)
20:10 pothos_ joined
tadzik ;) 20:10
20:11 pothos_ is now known as pothos, mj41 joined 20:12 y3llow left, packetknife left
dalek kudo/optimizer: 5b40d01 | moritz++ | src/core/ (4 files):
use native types in the setting a bit more
 20:13
20:13 y3llow joined, GlitchMr left, pothos left, pothos_ joined 20:14 pothos_ is now known as pothos
flussence nom: say ('ring' X '-' xx 6, '… bananaphone!').join 20:14
p6eval nom ff78b5: OUTPUT«ring-ring-ring-ring-ring-ring-ring… bananaphone!␤»
20:14 wamba left 20:15 y3llow left, y3llow joined
masak nom: given {}.^name -> $_ is copy { .=uc; .=subst('H', 'M'); $_ = .comb.join(' * '); .say } 20:16
p6eval nom ff78b5: OUTPUT«M * A * S * H␤»
flussence nom: print 'ring' X '-' xx 6, '… bananaphone!' # shorter 20:17
masak tadzik: new one ;)
p6eval nom ff78b5: OUTPUT«ring-ring-ring-ring-ring-ring-ring… bananaphone!»
tadzik masak: haha, nice :)
masak TimToady: sometimes I'm wondering if .=rhs shouldn't be a macro instead, expanding to $_.=rhs, with rhs possibly being several chained method calls; see above.
20:18 mattp_ left
flussence I want a call-a-bunch-of-methods shorthand like ». is for variables... 20:18
tadzik given? 20:19
I don't think I understand
oh, i think I understand
flussence you can write ($a, $b, $c)».foo, but not $a.«(<foo bar etc>) 20:20
20:20 mattp_ joined
masak flussence: a bit worrying that in your proposed syntax, the method names are strings... 20:20
flussence yeah, not sure about the details.
masak std: my $a; $a.«(<foo bar etc>) 20:21
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Unable to parse quote-words subscript; couldn't find right double-angle quote at /tmp/v4Lx0OvsVs line 1 (EOF):␤------> my $a; $a.«(<foo bar etc>)⏏<EOL>␤Parse failed␤FAILED 00:01 120m␤»
masak std: my $a; $a«.(<foo bar etc>)
p6eval std 20ae3bd: OUTPUT«===SORRY!===␤Unable to parse quote-words subscript; couldn't find right double-angle quote at /tmp/3WegW9TvCg line 1 (EOF):␤------> my $a; $a«.(<foo bar etc>)⏏<EOL>␤Parse failed␤FAILED 00:01 120m␤»
sorear good * #perl6
phenny sorear: 13:13Z <[Coke]> tell sorear that when I fudge roast for niecza, he should definitely check to make sure he's happy with the new tests.
flussence and both of them look like the start of a hash subscript... 20:22
masak hola, sorear
sjohnson nom: "efbc88e38080efbd80e383bcc2b4efbc890a".pack("H*").print 20:25
p6eval nom ff78b5: OUTPUT«Method 'pack' not found for invocant of class 'Str'␤ in block <anon> at /tmp/cL0ao0RmEu:1␤ in <anon> at /tmp/cL0ao0RmEu:1␤»
masak I'll get to it, I promise :) 20:26
but even then, you'd have problems with the Str/Buf distinction above.
I can never quite remember which way without looking... 20:27
flussence nom: my $a = 1; {.sin, .cos, .tan}».map({$a.$_})
p6eval nom ff78b5: ( no output )
flussence oh wait
nom: my $a = 1; say {.sin, .cos, .tan}».map({$a.$_})
p6eval nom ff78b5: OUTPUT«0.841470984807897 0.54030230586814 1.5574077246549␤»
sorear backlog complete
masak: you can take your hat off now.
flussence it's close, but backwards and messy
sjohnson what's the pack equivalent in perl 6? 20:28
sorear diakopter: in 'say: say 5', say IS allowed as a label, but it effectively shadows CORE::say
niecza: say: &say(4); goto say; # using &say explicitly disambiguates 20:29
p6eval niecza v10-38-g23f1b2e: OUTPUT«===SORRY!===␤␤Unsupported use of bare 'say'; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument at /tmp/wFcMxL_aNS line 1:␤------> say: &say(4); goto say⏏; # using &say explicitly disambiguat…
flussence std: goto 'say'
p6eval std 20ae3bd: OUTPUT«ok 00:01 120m␤»
sorear niecza: say: &say(4); goto 'say'; # using &say explicitly disambiguates
tadzik sjohnson: it's "pack", but NYI in nom
p6eval niecza v10-38-g23f1b2e: OUTPUT«(timeout)4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4􏿽xE2􏿽x90
sjohnson o
tadzik b: "efbc88e38080efbd80e383bcc2b4efbc890a".pack("H*").print
p6eval b 1b7dd1: OUTPUT«Method 'pack' not found for invocant of class 'Str'␤ in main program body at line 22:/tmp/bIFCI1d64Y␤»
masak sorear: seems I'm not authorized to take my hat off. moritz op'd me.
tadzik eek 20:30
sorear niecza: say: CORE::say 4; goto 'say'; # this works too diakopter
flussence std: 10: print "HELLO WORLD\n"; 20: goto 10
sorear masak: /mode #perl6 -o masak ?
p6eval niecza v10-38-g23f1b2e: OUTPUT«(timeout)4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4␤4􏿽xE2􏿽x90
std 20ae3bd: OUTPUT«===SORRY!===␤Illegal use of colon as invocant marker at /tmp/ZVkIMA2_Lw line 1:␤------> 10:⏏ print "HELLO WORLD\n"; 20: goto 10␤Check failed␤FAILED 00:01 122m␤»
20:30 MayDaniel left, masak sets mode: -o masak
tadzik :) 20:30
masak ooh :)
PerlJam flussence: remember, all's fair if you predeclare. you didn't first load BASIC::Grammar :)
masak sorear++ 20:31
20:31 phenny left, mj41 left
sorear masak: what were you trying? 20:31
masak: btw, I just gave you a chanserv can-use-the-op-command bit 20:32
masak thanks.
I was trying /msg Chanserv deop #perl6 masak
moritz /msg chanserv op #perl6
is the magical incantation then
and you can deop yourself
masak not with ChanServ I couldn't...
moritz well, with 'yourself' I meant /deop masak :-) 20:33
masak ok :)
sorear pm, moritz, and I are the metaops here. 20:34
sjohnson rakudo: "efbc88e38080efbd80e383bcc2b4efbc89".pack("H*").print
p6eval rakudo ff78b5: OUTPUT«Method 'pack' not found for invocant of class 'Str'␤ in block <anon> at /tmp/nPy85Y4V11:1␤ in <anon> at /tmp/nPy85Y4V11:1␤»
sjohnson i guess since nom won't do it, i shouldn't have bothered with that. oops! 20:35
masak sjohnson: .pack would be a method on List/Parcel, but I guess there could be a delegator on Cool as well.
tadzik time for fat-face-with-open-eyes! :)
20:35 wolfman2000 left
sjohnson （　° ー°） 20:35
tadzik :D
I have to copypaste those somewhere 20:36
sjohnson The fatface monologues
masak lol
flussence this talk of .pack reminded me of a thing I wanted to do but never got around to... gist.github.com/1186468 20:38
sorear 11:23 < masak> sorear: how far along are you with the /serialization branch?
I'd say I'm around half done; probably a bit more than half when it comes to hard stuff, a bit less for yakshavy bits 20:39
ETA 1-2 weeks, depending on tuit availability
11:23 < masak> sorear: and what will it mean, exactly, when it lands?
moritz instant world domination 20:40
proper role composition
masak as if Niecza needed more of that... :P
world domination, I mean.
moritz it does
im2ee Good night! :)
masak 'branoc! 20:41
sorear the immediate effects will be slower compilation, slightly faster startup, and more availability of metastuff at runtime
post-merge, I'll tackle roles and BEGIN, both of which are clearly blocking on this
masak *nod*
why are roles blocking on it?
sorear because I don't want to implement role composition twice 20:42
we need the ability to compose roles at compile time and at runtime
in niecza/master, those are two completely different metamodels
moritz I've just did an experiment; I tried to think of all the corner cases of the &min function, and implement it 20:45
gist.github.com/1265646
looks surprisingly lengthy/complex -- is there anything I can significantly simplify?
20:45 hamza joined
masak # FIRST next; 20:46
nice!
moritz++ # idiom discovery 20:47
tadzik what does it do?
besides "nothing, it's commented out"?
like "start from the second iteration"? 20:48
moritz tadzik: skip the first iteration, if not commented out
right
sorear moritz: don't you also need special cases for handling Ranges?
moritz sorear: care to elaborate?
sorear I vaguely thought min was overloaded to do interval operations 20:49
20:49 phenny joined
masak moritz: you don't need the trinary at L15 if you just assign -Int at L5. 20:49
er, -Inf.
perl6: say -Int
p6eval rakudo ff78b5: OUTPUT«Cannot unbox type object to a native integer␤ in sub prefix:<-> at src/gen/CORE.setting:2259␤ in block <anon> at /tmp/trrLYZoBzd:1␤ in <anon> at /tmp/trrLYZoBzd:1␤»
..pugs, niecza v10-38-g23f1b2e: OUTPUT«0␤»
moritz sorear: minmax does some weird Range stuff, but I'm not aware of any in min
masak submits rakudobug
sorear moritz: you can combine the two clauses if you allow both a $comparer and a $transformer
masak it would be good to have it be one code path, not two similar-ish ones. 20:51
20:51 im2ee left
moritz sorear: you're right; separating the loops is probably premature optimization 20:51
20:51 hamza left
sorear moritz: if you really want to deoptimize, &cmp = { &by($^a) cmp &by($^b) } 20:52
and what's the &by.arity == 0 case good for? 20:53
moritz sorear: :by(&rand)
sorear there is no &rand
PerlJam moritz++
moritz sorear: :by(-> { rand }) then
sorear ...why would you do that?
moritz sorear: people write code like that. If you don't guard against the zero-arity case, masak will write a bug report 20:54
PerlJam heh
:by( -> { -1 }) is more likely I think
20:55 mj41 joined
moritz .[0] would be cheaper 20:57
masak sorear: that's true, I will.
sjohnson any one here an "flock master" ? 20:58
benabik sjohnson: I just had do deal with some problems with it, but I'm not a master exactly. 20:59
sjohnson benabik: cool. it's about time i finally got proficient with the basics of it 21:00
PerlJam Is a flock master the lead duck?
sjohnson heh
masak does it count if one has player some "Angry Birds"?
played*
sjohnson less that, more a guy who understands flock() like the back of his hand
masak: i suppose it could :)
benabik sjohnson: Ah. I now know enough to do some simple debugging involving NFS. 21:01
moritz wonders what kind of understanding the back of a hand needs
PerlJam sjohnson: I've pondered flock more than I've pondered the back of my hand I think.
benabik ( flock -s 200 || exit 1; echo locked! ) 200>mylock
sjohnson if a script locks a text file... and some other script that doesn't use any flock statements tries to modify it... will the changes be allowed?
benabik sjohnson: That's the kind of thing I don't know. :-D 21:02
sjohnson in other words, does flock() go by the honesty policy?
moritz yes
ie, yes, changes allowed
PerlJam sjohnson: yes
sjohnson: unless it's on a system that requires mandatory locking
er .. you know what I mean 21:03
sjohnson interesting. so, if say that 2nd script wanted to do the right thing, would you need some sort of explicit flock() statement that requests to use it, and thus would be rejected since the 1st script used it
?
PerlJam s/requires/only supports/ # now you really know what I meant :)
moritz sjohnson: yes
flussence if you're trying to write a file safely, the safe way is to write to a temp copy, fsync then rename over the old version. Modifying in-place is bad. 21:04
masak even utilities that modify in-place do what flussence suggested :)
21:04 kaare_ left
flussence (a lot of people found this out the hard way when ext4 became widespread) 21:05
PerlJam flussence: er ... why? What did ext4 have to do with it?
moritz sjohnson: fwiw you can find out about the locking stuff quite nicely by doing experiments
sjohnson: in one console, write
perl -Mautodie -wE 'open my $h, "README"; flock $h, 2; sleep(5000)'
sjohnson yeah, i was about to bust open my classic "farm".
sjohnson% touch pig cow chicken 21:06
flussence PerlJam: they changed the default journal mode, so rename-then-fsync went horribly wrong or something like that
moritz and in another one, try echo 'foo' > README
sjohnson moritz: i like your idea better than mine!
PerlJam sjohnson: Is this on linux btw?
sjohnson yeah
just some ubuntu shell i have
flussence (I think a workaround's been added to ext4 since then) 21:07
PerlJam sjohnson: if you have root, you can remount a partition for mandatory locking and see what the difference is
sjohnson learns much today 21:08
PerlJam sjohnson: some systems, (like Windows I believe) only have mandatory locking, so if you're writing for portability you should understand what happens :) 21:09
moritz masak: why do you think that -Int should not warn?
sjohnson i suppose i might understand more with this too: en.wikipedia.org/wiki/File_locking
21:09 ggoebel left
sjohnson PerlJam: good to know. thankfully, no portability with this particular script.. but good to keep in mind 21:09
thanks for the help 21:10
*p6 group hug*
masak moritz: because I can see it being part of a legitimate mathematical computation, where it would make more sense for it to pun to 0 than to warn. 21:11
moritz: I think we have several of these cases.
PerlJam sjohnson: so ... when will you have flock implemented for Perl 6? :-)
sjohnson if i tried, i'm sure it would be the collapse of the internet 21:12
i have 0 faith in my low-level programming skills at this present time
moritz masak: note that ++ and -- and *= etc. don't warn
sjohnson especially if some bank used rakudo
moritz nom: +Int
p6eval nom ff78b5: OUTPUT«Use of uninitialized value in numeric context␤»
sjohnson might get sued millions of dollars
sorear moritz: I think it would more more correct to die(..) unless &by.arity == 1|2 21:13
masak moritz: oh, so maybe it's just the "mutating" math operators that don't warn? 21:14
moritz masak: yes
masak I'll add that to the ticket (and change the subject)
dalek kudo/nom: d3e6519 | moritz++ | src/core/Int.pm:
add some missing :D markers on Int ops, masak++
21:15 bluescreen100 left, bluescreen10 left
jnthn moritz: That...probably just hosed inlining :( 21:15
sjohnson how hard is it to understand parrot programming language?
sorear masak: per spec, $x op= $y means something like $x = ($x // [op]) op $y
moritz sjohnson: there is no "parrot programming language"
jnthn: then inlining needs to improve wrt D 21:16
masak sjohnson: did you mean PIR? nqp?
moritz erm, :D
masak sjohnson: C? :)
sjohnson PIR
i would imagine that rakudo is written in some sort of parrot-script, if you will
moritz jnthn: those :D's aren't in there for fun, but because we need them
masak sjohnson: no, not anymore. 21:17
sjohnson wonders where he's been all this time
masak sjohnson: Rakudo is written in Perl 6, nqp, and some C.
jnthn moritz: It's kinda trickyish.
sjohnson what's NQP?
not quite perl.. hmm
21:17 dual left
moritz a subset of Perl 6 21:17
jnthn moritz: We have to know try and track declaredness as well as type.
sjohnson tries to understand
tadzik hehe
"add some missing :D" -- moritz
sjohnson so, if you want to compile rakudo, what gets compiled first? C? 21:18
PerlJam sjohnson: NQP is just enough Perl 6 to implement Perl 6 :)
tadzik sjohnson: yes
sjohnson trying to figure what the first stage is of perl6 being interpretted to build rakudo
tadzik then the compiler, which is nqp
then the setting, which is Perl 6
sjohnson perl6 dev team has been busy bees!
moritz jnthn: maybe we need to declare variables as Int:D more often 21:19
... or maybe that should become the default, just like in params
jnthn moritz: Yeah, we don't implement that defaulting yet.
moritz: Though it shouldn't be hard to.
tadzik :D is default in params?
jnthn moritz: There's only one problem with it for variables 21:20
my Str $s;
moritz right
PerlJam tadzik: :D == defined
tadzik yes, I know
sorear :D is default in params?
moritz jnthn: that might need to assign a default
PerlJam oh.
jnthn sorear: Apparently. ;)
tadzik huh
sorear when was this decided?
PerlJam is only paying about 0.25 attention to the N channels he's looking at on IRC
moritz or maybe it's only on the invocant?
moritz checks the spec
jnthn oh, maybe...I thought it was done for params generally 21:21
flussence rakudo: sub x(Str $x) { say $x.WHAT }; x(Str)
p6eval rakudo ff78b5: OUTPUT«Str()␤»
moritz the commit log of a4e255cdec866af833f91d1e3c0d9ce6ddbdf74f says "Invocants now default to :D"
jnthn oh
Hm, ok
That's less happy for the inliner :) 21:22
moritz huh, I thought the :D caused problem for the inliner?
PerlJam jnthn: is there some preference between :D and :U in MMD?
moritz totally confused
masak sorear: I remember this being talked about during pre-YAPC::EU.
jnthn moritz: Yes and no (more) 21:23
moritz: Needing to know whether something is defined or not to inline is only a problem if the common case is that we don't easily know.
moritz: If stuff defaults to :D and you have to declare otherwise, the optimizer knows plenty.
If the situation is "we need to know it's :D to inline the candidate, and we have to infer that", then it's a little tricky. 21:26
moritz right
gfldex nom: my Str $foo = '12.345'; say $foo.Rat;
p6eval nom ff78b5: OUTPUT«Method 'Rat' not found for invocant of class 'Str'␤ in block <anon> at /tmp/ERpEWJVnpT:1␤ in <anon> at /tmp/ERpEWJVnpT:1␤»
moritz nom: say '12.234'.Numeric.WHAT 21:27
p6eval nom ff78b5: OUTPUT«Rat()␤»
21:28 dual joined
moritz conducts an experiment 21:29
masak moritz: hey, what's this about giving me tickets for testing? :P 21:31
sorear masak: eh, so I just need to get to YAPC::EU someday. ok.
moritz masak: the experiment :-) 21:32
masak my interests lie more in the area of increasing the RT queue size...
sorear: yes, but not for that reason. that one discussion already happened.
moritz an attempt to suck you into the test suite :-)
a mini challenge. Don't you love mini challenges? :-) 21:33
masak :P
moritz: I admire your tenacity wrt the test suite. I occasionally make dips into it, but I find I cannot sustain maintaining it for some reason. 21:34
I can sustain submitting rakudobugs.
dalek kudo/nom: fb53581 | moritz++ | src/core/Cool.pm:
missing coercers, gfldex++
 21:35
moritz cannot sustain to stay awake, and thus fades to sleep
21:35 cognominal left, cognominal joined
gfldex what could "Can only use get_what on a SixModelObject" mean? 21:36
masak moritz: dream of dozens of autonomous moritz agents, living on the 'Net and tending the test suite.
gfldex: that you've somehow dropped out of the normal type system.
gfldex: my guess is metamodel, or different HLL.
benabik Likely a raw parrot object leaked out somewhere. 21:37
sorear or doing unspeakable things with pir::
gfldex i would never even try to do unspeakable things with pir::
jnthn metamodel is all .WHATable
gfldex say %.components.WHAT; 21:38
that's the line causing it
aint have no pir:: in it
jnthn Where is that line?
In a method?
gfldex yes
benabik %.components? Shouldn't there be something between % and . ? 21:39
jnthn gfldex: OK, I'll need a bit more context. If you can golf it down to something small, I can investigate...
benabik (Although I don't think it should leak parrot into the system either.
gfldex looks for his golf clubs 21:40
masak benabik: '%.components' looks fine to me. 21:41
benabik masak: What does it mean?
masak benabik: it means 'public hash attribute'
sorear benabik: %.components is shorthand for self.components.hash
masak benabik: the '.' is a twigil.
benabik Oh. *facepalm* 21:42
gfldex i'm trying to nomify a script that used to work with master
benabik I forgot . could be a twigil too. *sigh*
dalek ast: fc78e0f | masak++ | S03-operators/arith.t:
added test for #100768

Because moritz++ told me to.		 21:49
21:55 benabik left 21:58 mj41 left 22:02 f00li5h joined 22:04 S11001001 joined
masak 'night, #perl6. dream of implementations with all green in perl6.org/compilers/features 22:09
sorear o/ S11001001 22:22
S11001001 hi sorear
22:27 [Coke] left
sjohnson heh 22:29
22:29 [Coke] joined
sjohnson is that binary for 0xC9 ? 22:29
S11001001 sjohnson: indeed
sjohnson oh, i didn't even see you there 22:30
i thought it was his way of saying: o/ 0b101010
iow, *
sorear what's the significance of 0xC9 22:35
22:36 am0c left
S11001001 just a star trek thing 22:37
22:37 kfo left 22:38 kfo joined
sjohnson sorear: do you use / like C at all? 22:40
sorear sjohnson: yes
sjohnson yes to both? :) 22:41
sorear yes
22:42 alvis left
sjohnson cool 22:43
another question
do you know much about Programming Pearls?
i wanted to beef up on some computer science knowledge, and there's that book.. and the TAOCP.. which might be too expensive, too difficult, and to large of a read 22:44
having a hard time deciding which avenue to go down.
dalek ecza/serialize: 11c6923 | sorear++ | / (6 files):
Reimplement sub "extend" functionality
 22:47
ecza/serialize: 63597b7 | sorear++ | / (4 files):
Reimplement return-pass, Whatever currying
sorear sjohnson: ALL OF THEM. :D 22:48
sjohnson not a bad idea
sorear sjohnson: I found TAOCP in a library. It still contained a few things I didn't already know
sjohnson that book's so big, probably would warrant me buying it 22:50
sorear other things you ought to read: *SICP *Draft Report on the EDVAC (to be continued)
sjohnson thanks 22:54
no idea what it means from the title
but sounds interesting nonetheless
22:55 alvis joined
TimToady google is your friend, at least this time 22:55
22:55 S11001001 left
sorear the latter was a document written around '45 that outlined the design of the modern computer 22:55
it's still useful because it was written for an audience without much computer experience :D 22:56
cognominal nom: 'aaaa' ~~ m/ $<a>=.*/; $/<a>.perl
phenny cognominal: 30 Sep 23:48Z <[Coke]> ask cognominal to update RT#75850 with the expected output.
p6eval nom fb5358: ( no output )
sjohnson i'd like to learn practical stuff like which data structures are good for what scenarios
instead of marvel over cool abrasct computer concepts 22:57
mostly casue i want to write an encryption app for free
cognominal nom: 'aaaa' ~~ m/ $<a>=.*/; say $/<a>.perl
p6eval nom fb5358: OUTPUT«Match.perl(orig => "aaaa", from => 0, to => 4, ast => Mu, list => ().list, hash => EnumMap.new())␤»
sjohnson already owns applied cryptography
sorear I wouldn't trust a cryptography app unless it was written by someone famous
there is a LONG history of independant cryptography apps that are actually completely useless because of a misunderstanding or something 22:58
sjohnson i am well aware of that, to be honest.
i was hoping to pay a professianal to have a conversation with me
so i could describe exactly what the protocl is
protocol is 22:59
and let me know if i'm a fool or not.
obviously i'd be for free, so no one would kill me for selling a faulty product at least.
sorear what do you want to do that gpg doesn't already do? 23:00
sjohnson gpg doesn't have perfect forward secrecy 23:07
also, i want to make a clone of Secway Simp, which encrypts IM traffic
like MSN, Jabber, ICQ, etc.
sorear erm, what do you mean perfect forward secrecy 23:09
23:11 hacktor joined
sjohnson private key compromised 23:11
means all previous messages that you can assume were sniffed and stored somewhere are readable
use a private key only for authentication, and use a diffie hellman to agree on a shared sym. key, will mean you need to solve the DH problem first, even if the private key is compromised 23:12
only thing an attacker could do with a private key once its compromised, is pretend to be you in future conversations, and not read past ones.
dalek kudo/optimizer: a47b0cd | jnthn++ | src/Perl6/Actions.pm:
Tighten up what we admit for inlining; require that the arguments are *all* mentioned.
 23:13
kudo/optimizer: fb111db | jnthn++ | / (3 files):
Add support for --optimize=[off|1|2|3]. The default is level 2. 'off' will not even enter the optimizer, so it skips CHECK time could-never-work analysis. Classify inlining and compile-time multi selection as level 2 for now; block inlining gets level 3 initially but should drop to two in the not too distant future; it needs some more aggressive testing. We build the setting with --optimize=3.
23:14 hacktor is now known as gttg
sjohnson sorear: see en.wikipedia.org/wiki/Perfect_forward_secrecy . also, Secway Simp deosn't even have PFS with their authenticated mode, they admitted in an email to me. 23:14
so my free product would probably the first one to do it. problem is, i need some better understanding of how things are done in the C programming world as well as algorithms as far as data storage go, so I don't look like the village idiot in the crypto community 23:15
jlaire < sjohnson> i'd like to learn practical stuff like which data structures are good for what scenarios
I recommend reading "Introduction to Algorithm" and solving lots of algorithm problems
sorear sjohnson: if you want to learn about the science of algorithms and data structures, READ TAOCP.
that's the main focus of the book series 23:16
jlaire is currently reading TAOCP and likin' it
sorear sjohnson: also read anything else you can find. Ask questions here, I collect obscure algorithms and data structures. 23:17
sjohnson thanks for the advice
sorear: so TAOCP isn't 100% esoteric? 23:18
i can apply that understanding to real world stuff?
jlaire yes
sjohnson sweet
will do!
sorear sjohnson: *everything* in TAOCP is usable in the real world.
sjohnson i'll just go ahead and get the goods then 23:19
sorear sjohnson: there are some books that are designed to be usable by people too busy to really learn how to program. TAOCP is not one of them
sjohnson uh oh. 23:20
i mean, i'm willing to put in time
but i don't know if i can devote all my free time for the next 15 years to trying to make sense of TAOCP
jlaire familiarity with assembly programming for machines from the '60s is especially useful ;)
sorear sjohnson: I'd say I've devoted 10% of my free time for the last 10 years to becoming a better programmer, and I had no troble reading TAOCP. 23:22
sjohnson oh okay
i should just read it, and not assume it's going to be the most impossible thing in the world. 23:23
there was once a time i thought it would be impossible for me to learn Perl
and i stuck with that distorted view for 4 years, and avoided it whenever a perl task came up at work. then i had to learn it, bought the camel book, and was doing some impressive stuff only a few days later
it was a great feeling, btw.
good life lesson 23:24
jlaire sorear: "It still contained a few things I didn't already know" <-- was "a few things" a slight exaggeration?
sorear jlaire: do you mean understatement?
jlaire urd
yes* 23:25
sorear jlaire: I'd say I already knew most of the really important things there, because of how long I've spent as a practicing programmer and especially how much reading I've done on algorithms
jlaire sorear: ok, I see. how much time did you spend on the exerices v. reading the book? 23:26
sorear 2:3 or so. I didn't systematically do all the exercises 23:27
jlaire tyvm
I don't thnk anyone could do them all ;) 23:28
sjohnson sorear: does it deal with big O notation, for someone like me who is unable to prove notations in a CS class?
or does it just assume you already know that, and skip it entirely
(taocp that is)
i'm not very familiar with some of the basic things, and i kind of need to start small.
sorear sjohnson: it talks about what O(n) is, yes
jlaire in that case, I'd recommend Concrete Mathematics, too 23:29
sjohnson cool
i've logged both your recommednations, by the way 23:30
so, it's not falling on deaf ears :)
sorear sjohnson: I'm a crypto geek too, I'd be interested in seeing your designs
jlaire TAOCP has a fairly lengthy section on mathematics, but I found it a bit terse
:)
TimToady O() is mostly just a funny way to talk about dimensionality
jlaire o_O
TimToady: sounds interesting, can you elaborate 23:31
sjohnson sorear: i had one change made in GnuPG 1.4.x branch, so at least i finally contributed to something
took a lot of arguing on the mailing list
TimToady it's about the size of search spaces
sorear the important thing with O() is that you understand compositionality
few mortals can prove that a lookup in a distinct-set forest takes O(A^-1(N)) time 23:32
TimToady O(1) basically means you have to move in 0 dimensions to achieve a solution
sjohnson but the security of DH with using a private key to authenticate only gives me hope for a better future of safety :) but GnuPG is pretty cool, i do like it a lot.
TimToady O(n) means you have to move in 1 dimension
O(n^2) means 2 dimensions, etc
sorear but, given that, every programmer needs to be able to intuit the time complexity of doing N lookups
jlaire
.oO( moving in fractal dimensions )
TimToady relationships between dimensions produce fractal dimensionality
sorear wonders how many dimensions there are in O(A^-1(N)) 23:33
sjohnson
.oO( ... ) is Big Think notation
jlaire sorear: what's a "distinct-set forest"? google gives no hits
sjohnson i also need to remember what log cruves sort of look like too.
curves* 23:34
TimToady they keep getting flatter, but never go flat
just like exponentials never quite go straight up
23:35 gttg left
sorear en.wikipedia.org/wiki/Disjoint-set_...et_forests 23:36
jlaire oh, disjoint makes more sense
sorear sjohnson: a log curve is one where going from 1 to 10 is 1 unit, and so is going from 1000 to 10000 23:37
TimToady fractal branches are generally disjoint
sjohnson i wonder how much of this stuff that facebook guy knew, before he made billions 23:38
i wonder if there's a point where one knows too much math
jlaire was told that the lookup is "practically amortized constant" when learning about this data structure 23:39
jnthn wonders if there's a point where on feels they know too many anything :)
*one
Well, there's probably some things I'd rather not know about but...there's always something interesting more to learn. :)
jlaire technical skills probably weren't the reason he made billions
TimToady wonders whether knowing both too little and too much about something are really mutually exclusive... 23:40
jnthn :) 23:41
sorear I think making billions is mostly a function of being evil
TimToady neither the converse nor the inverse follow, of course :) 23:42
dalek kudo/optimizer: c191cfe | jnthn++ | src/Perl6/Optimizer.pm:
Improve error reporting. When we determine a call could never work, shows the desired signature(s) in the error.
kudo/optimizer: b5d2c9b | jnthn++ | src/Perl6/Actions.pm:
Try to get better line reporting for calls.
 23:45
23:46 jaldhar joined 23:50 itz left 23:53 cggoebel joined 23:57 itz joined 23:59 wolfman2000 joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!