»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.
00:04 LlamaRider left 00:05 davidagain_ left 00:29 japhb_ left 00:38 Liz_ joined, Liz_ is now known as lizmat 00:39 Chillance left
timotimo what would be a performant way to have a map of tiles that are either occupied or not and i want to set and check single tiles 00:40
at the moment i have a hash of "$x|$y" strings as keys and bools as values. that doesn't seem optimal
diakopter how big can the dimensions get 00:41
timotimo like rosettacode.org/wiki/Brownian_tree - trying to implement this, the other implementations have 400x400 00:42
diakopter a 400-item array of 50-byte bufs where each bit is a cell? 00:43
er, whatever the mutable int array is 00:44
timotimo isn't that a bit unwieldy?
don't really want to do bitwise operations like shifts and masks
diakopter a byte unwieldy, maybe
timotimo :D
diakopter then just use a 400 item array of 400 int arrays :P 00:45
I was merely suggesting saving space
I mean, 160,000 * 8 bytes isn't too huge. if your dimensions were 40000x40000, on the other hand... 00:47
oh, and, your mom is a bit unwieldy. not to mention she's a Brownian tree. 00:48
.. that was a silly joke, for anyone who might've thought otherwise. 00:49
.. that it wasn't a joke, not that it wasn't silly.. *facepalm* 00:51
timotimo i've tried that, but i can't assign to it :(
diakopter eh?
timotimo it works with :=, though
doesn't make it very much faster. weird. 00:52
diakopter r: my Array[Array[int]] @a; @a[399][399] := 1;
p6eval rakudo 1c9939: OUTPUT«Cannot look up attributes in a type object␤ in method gimme at src/gen/CORE.setting:5897␤ in method bind_pos at src/gen/CORE.setting:6351␤ in method postcircumfix:<[ ]> at src/gen/CORE.setting:1407␤ in block at /tmp/chpr3el86Z:1␤␤»
timotimo at least for niecza
diakopter erm
n: my Array[Array[Int]] @a; @a[399][399] := 1; 00:53
p6eval niecza v24-34-g4f8f894: OUTPUT«Unhandled exception: Cannot use hash access on an object of type Array␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.setting line 355 (Any.bind_key @ 9) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at <unknown> line 0 (KERNEL An…
diakopter erm
diakopter fails at life
00:53 toddr_ left
timotimo rn: my @a; @a[$_] = 0 xx 400 for ^400; @a[50][50] = 1; 00:54
p6eval rakudo 1c9939: OUTPUT«(timeout)»
..niecza v24-34-g4f8f894: OUTPUT«Unhandled exception: Writing to readonly scalar␤ at /tmp/qVg2AtKmzV line 1 (mainline @ 9) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4299 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4300 (module-CORE @ 583) ␤ at /home/p6eval/niecza/li…
timotimo rn: my @a; @a[$_] = 0 xx 400 for ^400; @a[50][50] := 1;
p6eval rakudo 1c9939: OUTPUT«(timeout)» 00:55
..niecza v24-34-g4f8f894: ( no output )
timotimo >_<
rn: my @a; @a[$_] = 0 xx 100 for ^100; @a[50][50] := 1;
diakopter lolz.
p6eval niecza v24-34-g4f8f894: ( no output )
..rakudo 1c9939: OUTPUT«No such method 'bind_pos' for invocant of type 'List'␤ in method postcircumfix:<[ ]> at src/gen/CORE.setting:1407␤ in block at /tmp/fPrFzkVlDn:1␤␤»
timotimo rn: my @a; @a[$_] = 0 xx 100 for ^100; @a[50][50] = 1;
p6eval rakudo 1c9939: OUTPUT«Cannot modify an immutable value␤ in block at /tmp/xGv4xqqChJ:1␤␤»
..niecza v24-34-g4f8f894: OUTPUT«Unhandled exception: Writing to readonly scalar␤ at /tmp/zUSheFaZnl line 1 (mainline @ 9) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4299 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4300 (module-CORE @ 583) ␤ at /home/p6eval/niecza/li…
diakopter we can't win.
timotimo nope, can't
diakopter p: my @a; @a[399][399] := 1; say @a[399][399] 00:56
p6eval pugs: OUTPUT«1␤»
timotimo mhm
diakopter p: my @a; @a[399][399] := 1; say @a[399] 00:57
p6eval pugs: OUTPUT« …
diakopter squints
timotimo it's too small!
diakopter p: my @a; @a[399][399] := 1; say @a[399].perl
p6eval pugs: OUTPUT«\\(undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef, undef…
diakopter your mom is undef
timotimo that is a non-insult 00:58
diakopter r: my @a; @a[399][399] = 1; say @a[399][399]
p6eval rakudo 1c9939: OUTPUT«1␤»
timotimo r: oh? interesting!
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/DuPXRyW1XP:1␤------> oh⏏? interesting!␤ expecting any of:␤ argument list␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement…
timotimo ... :D
diakopter r: term term
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Undeclared routine:␤ term used at line 1␤␤» 00:59
diakopter r: .term .term
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/CAfb1PgmuN:1␤------> .term ⏏.term␤ expecting any of:␤ method arguments␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement m…
diakopter there
timotimo :D
diakopter r: my @a; @a[0][1][2][3][4][5][6][7][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say @a 01:01
p6eval rakudo 1c9939: OUTPUT«use of uninitialized value of type Any in string context in block at /tmp/1ejPZTUhEN:1␤␤use of uninitialized value of type Any in string context in block at /tmp/1ejPZTUhEN:1␤␤use of uninitialized value of type Any in string context in block at /tmp/1ejPZTUhE…
diakopter r: my @a; @a[0][1][2][3][4][5][6][7] = 1; say @a
p6eval rakudo 1c9939: OUTPUT«use of uninitialized value of type Any in string context in block at /tmp/IHqmZy4XWp:1␤␤use of uninitialized value of type Any in string context in block at /tmp/IHqmZy4XWp:1␤␤use of uninitialized value of type Any in string context in block at /tmp/IHqmZy4XW… 01:02
diakopter *sigh*
timotimo p: my @a; @a[0][1][2][3][4][5][6][7] = 1; say @a
p6eval pugs: OUTPUT« 1␤»
TimToady r: my @a = [0 xx 400] xx 400; say @a[399][399].WHAT
colomon n: my @a; @a[0][1][2][3][4][5][6][7] = 1; say @a
p6eval rakudo 1c9939: OUTPUT«(timeout)»
niecza v24-34-g4f8f894: OUTPUT«(Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) (Any) 1␤»
TimToady n: my @a = [0 xx 400] xx 400; say @a[399][399].WHAT 01:03
p6eval niecza v24-34-g4f8f894: OUTPUT«(Int)␤»
diakopter r: my @a; @a[0][1][2][3][4][5][6][7] = 1; say @a.perl
p6eval rakudo 1c9939: OUTPUT«Array.new([Any, [Any, Any, [Any, Any, Any, [Any, Any, Any, Any, [Any, Any, Any, Any, Any, [Any, Any, Any, Any, Any, Any, [Any, Any, Any, Any, Any, Any, Any, 1]]]]]]])␤»
TimToady timotimo: ^^
timotimo using just @map and setting values on it un-pre-initialised will give me like 20 seconds speed-up
TimToady that works too 01:04
timotimo (still takes 180 seconds)
TimToady boolean tests work find on undefineds :)
*fine
diakopter r: my @a; @a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say @a.perl
p6eval rakudo 1c9939: OUTPUT«Array.new([[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[1]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]])␤»
diakopter hee
TimToady so why is only the outermost one Array.new() 01:05
diakopter r: my $a; $a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say $a.perl
p6eval rakudo 1c9939: OUTPUT«[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[1]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]␤»
diakopter no idea
01:06 am0c joined
diakopter r: my $a = []; $a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say $a.perl 01:06
p6eval rakudo 1c9939: OUTPUT«[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[1]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]␤»
TimToady trying to indicate flattenable, I guess
timotimo r: my $a; $a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; $a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][1][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say $a.perl
p6eval rakudo 1c9939: OUTPUT«[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[1]]]]]]]]]]]]]], [[[[[[[[[[[[[[1]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]␤»
timotimo no longer an array.new
diakopter r: my $a = (); $a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say $a.perl 01:07
p6eval rakudo 1c9939: OUTPUT«Cannot modify an immutable value␤ in sub infix:<=> at src/gen/CORE.setting:12590␤ in block at src/gen/CORE.setting:1533␤ in sub infix:<=> at src/gen/CORE.setting:12590␤ in block at src/gen/CORE.setting:1533␤ in sub infix:<=> at src/gen/CORE.setting:12590␤ i…
timotimo that does make some sense
diakopter r: my $a = \[]; $a[0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0][0] = 1; say $a.perl 01:08
p6eval rakudo 1c9939: OUTPUT«Capture.new( list => ([[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[1]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]],).list, hash => EnumMap.new())␤»
diakopter r: my $a = \$a; $a[0] = 1; say $a.perl
p6eval rakudo 1c9939: OUTPUT«Cannot assign to a readonly variable or a value␤ in block at /tmp/u7QSE5HcUT:1␤␤»
diakopter whatevs 01:09
r: my $a = \(); $a[0] = 1; say $a.perl 01:10
p6eval rakudo 1c9939: OUTPUT«Cannot modify an immutable value␤ in block at /tmp/hIlrbbcUnp:1␤␤»
diakopter r: my $a = \$a; say $a.perl # why does it timeout before recurse overflow 01:12
p6eval rakudo 1c9939: OUTPUT«(timeout)»
timotimo tail call optimisation? ;))
nope.
diakopter hm, I guess it could make its own stack 01:13
np: my $a = \$a; say $a.perl # haha
timotimo haha? 01:14
diakopter oh. apparently pn: and np: don't work
pn: die
p: my $a = \$a; say $a.perl 01:15
p6eval pugs: OUTPUT«\undef␤»
diakopter n: my $a = \$a; say $a.perl
p6eval niecza v24-34-g4f8f894: OUTPUT«(timeout)»
diakopter pugs++
timotimo so i'm not likely going to get something more performant without compromising code legibility
TimToady huh, I was gonna wait on rosettacode.org/wiki/24_game/Solve#Perl_6 till there was .permutations built-in, but I see thundergnat went ahead and did it anyway :D 01:17
diakopter I don't think a bit array [nicely abstracted] would make it less legible. also, I have no idea whether it would be more efficient.
TimToady my bit @a[400][400] is specced, just not implemented 01:18
diakopter how about unary represented in binary? 01:19
timotimo may be more efficient. may not be faster. who knows that now :(
n: my Array of Array of Int @map;
p6eval niecza v24-34-g4f8f894: OUTPUT«===SORRY!===␤␤Coercive declarations NYI at /tmp/QAFsg7zB8v line 1:␤------> my Array of Array of Int⏏ @map;␤␤Potential difficulties:␤ @map is declared but not used at /tmp/QAFsg7zB8v line 1:␤------> my Array of Arr…
timotimo :( 01:20
diakopter p: my unary $stuff; say $stuff.perl
p6eval pugs: OUTPUT«\unary()␤»
diakopter o_O
TimToady r: my $b = Buf.new(0 xx 40); $b[20] = 1; say $b
p6eval rakudo 1c9939: OUTPUT«Cannot modify an immutable value␤ in block at /tmp/xmBeqPnP00:1␤␤»
TimToady ooh, rakudobug
diakopter someone said Buf was immutable?
or was that buf
TimToady Bufs are supposed to be mutable
diakopter which one is fixed-size? 01:21
either?
r: my $b = Buf.new(0 xx 0); $b[20] = 1; say $b
p6eval rakudo 1c9939: OUTPUT«Cannot take substr outside string␤ in method at_pos at src/gen/CORE.setting:7358␤ in method postcircumfix:<[ ]> at src/gen/CORE.setting:1403␤ in block at /tmp/Qd9Af9yFAr:1␤␤» 01:22
diakopter ...
r: my $b = Buf.new(0 x 0); $b[20] = 1; say $b
p6eval rakudo 1c9939: OUTPUT«This type cannot unbox to a native integer␤ in method set_codes at src/gen/CORE.setting:7350␤ in method new at src/gen/CORE.setting:7342␤ in block at /tmp/orWNkDB0f_:1␤␤»
diakopter *headdesk*
r: my $b = Buf.new(-1); $b[20] = 1; say $b 01:23
p6eval rakudo 1c9939: OUTPUT«Invalid character for UTF-8 encoding␤␤ in method set_codes at src/gen/CORE.setting:7350␤ in method new at src/gen/CORE.setting:7342␤ in block at /tmp/BKa8CvABEx:1␤␤»
diakopter *headfloor*
TimToady UTF-8?!?! 01:24
diakopter tries to stop rotfl
geekosaur wonders if it's trying to print it as bytes or something 01:25
TimToady r: my $b = Buf.new(-1); 01:26
p6eval rakudo 1c9939: OUTPUT«Invalid character for UTF-8 encoding␤␤ in method set_codes at src/gen/CORE.setting:7350␤ in method new at src/gen/CORE.setting:7342␤ in block at /tmp/j_tvgAP6K5:1␤␤»
TimToady no
diakopter r: my $b = Buf.new(0-(2**63 - 1)); say $b.elems
p6eval rakudo 1c9939: OUTPUT«1␤»
diakopter gasps for air 01:27
TimToady r: my $b = Buf.new(255);
p6eval rakudo 1c9939: ( no output )
timotimo i don't know what's what any more
geekosaur eyes latest "bug" 01:34
diakopter oh. 01:35
the args to Buf.new are the elements..
ergh
timotimo yeah, i was confused about your confusion 01:36
diakopter r: my $a = []; $a[999999999] = 1; my $b = Buf.new(|$a); say $b.elems 01:37
p6eval rakudo 1c9939: OUTPUT«(signal SEGV)»
diakopter finally... a real bug
r: my $a = []; $a[999999999] = 1; my $b = Buf.new(|$a); 01:38
timotimo hehe
p6eval rakudo 1c9939: OUTPUT«(signal SEGV)»
timotimo so, craptons of zeros, then a one
and it'll break
diakopter r: my $a = []; $a[2**31 - 1] = 1; my $b = Buf.new(|$a);
p6eval rakudo 1c9939: OUTPUT«ResizablePMCArray: Can't resize!␤ in block at src/gen/CORE.setting:6335␤ in block at /tmp/r8IBzgCpKj:1␤␤» 01:39
diakopter oh dear.
I sense another fuzzing attack coming on.
TimToady well, Buf is really only about half implemented, so it's kinda unfair... 01:42
diakopter actually the Buf golfs away
01:44 lizmat left 01:45 Liz_ joined 01:59 sqirrel_ joined 02:02 sqirrel__ left
diakopter r: my $a = []; $a[240731133] = 1; 02:18
p6eval rakudo 1c9939: OUTPUT«(signal SEGV)»
diakopter r: my $a = []; $a[240731132] = 1;
golfed
p6eval rakudo 1c9939: OUTPUT«(timeout)»
diakopter has no clue about the significance of that threshold, and I'm too braindead to imagination anything 02:19
r: my Array[0] $a; 02:20
p6eval rakudo 1c9939: OUTPUT«Cannot look up attributes in a type object␤current instr.: 'gimme' pc 262278 (src/gen/CORE.setting.pir:119028) (src/gen/CORE.setting:5897)␤called from Sub 'sink' pc 270805 (src/gen/CORE.setting.pir:122461) (src/gen/CORE.setting:6180)␤called from Sub 'MAIN' pc 381 (…
diakopter ok that's just incomprehensible 02:21
sorear i would guess rakudo is allocating slightly more memory than it needs to, to avoid reallocations
and 240731133 is just the point where some internal variable goes negative 02:22
since the underlying objects in a QRPA are pointers, and p6eval runs on an I32LP64 machine
pmichaud TimToady: shall I submit the "Parallelism in Perl 6" talk to yapc::na? Want me to include your name on it?
sorear r: 2**31 / 8 # number of pointers needed to overflow an int 02:23
p6eval rakudo 1c9939: ( no output )
sorear r: say 2**31 / 8 # number of pointers needed to overflow an int
p6eval rakudo 1c9939: OUTPUT«268435456␤»
diakopter p: my $a = []; $a[240731133] = 1;
p6eval pugs: OUTPUT«Stack space overflow: current size 8388608 bytes.␤Use `+RTS -Ksize -RTS' to increase it.␤»
sorear pmichaud: isn't qrpa yours? can you comment on my theory? 02:24
pmichaud sorear: I wrote qrpa, yes... but a lot of it is cargo-culted from RPA
your theory is very sound to me, though.
it wouldn't surprise me if large arrays fail in bad ways at this point 02:25
TimToady pmichaud: yes, please, and yes, please, and thank you 02:26
pmichaud TimToady: will do, ty!
sorear: on the other hand, RPA doesn't seem to have a problem with that.
oh, wait, I might've entered it wrong
gist.github.com/pmichaud/5174660 02:28
diakopter p: my $a = []; $a[1016809] = 1;
p6eval pugs: OUTPUT«Stack space overflow: current size 8388608 bytes.␤Use `+RTS -Ksize -RTS' to increase it.␤»
diakopter p: my $a = []; $a[1016808] = 1; 02:29
p6eval pugs: ( no output )
diakopter doesn't know why it uses the stack
sorear: could ghc be recursing for each item? o_O 02:30
pmichaud nqp: my @a := []; @a[240731133] := 1; say('alive');
p6eval nqp: OUTPUT«(timeout)»
diakopter heh
pmichaud nqp: my @a := []; @a[20] := 1; say('alive');
p6eval nqp: OUTPUT«alive␤»
pmichaud nqp: my @a := []; @a[2407311] := 1; say('alive'); 02:31
p6eval nqp: OUTPUT«alive␤»
pmichaud nqp: my @a := []; @a[240731133] := 1; say('alive');
p6eval nqp: OUTPUT«(timeout)»
pmichaud we're probably pounding on poor feather[0-9] a fair bit :-/
diakopter nqp-jvm: my $a := []; $a[240731133] := 1; say('alive');
p6eval nqp-jvm: OUTPUT«java.lang.OutOfMemoryError: Java heap space␤ at org.perl6.nqp.sixmodel.reprs.VMArrayInstance.set_size_internal(VMArrayInstance.java:83)␤ at org.perl6.nqp.sixmodel.reprs.VMArrayInstance.bind_pos_boxed(VMArrayInstance.java:102)␤ at 02:32
..org.perl6.nqp.runtime.Ops.bindpos(Ops.jav…
diakopter nqp-jvm: my $a := []; $a[240731133000000000] := 1; say('alive');
p6eval nqp-jvm: OUTPUT«java.lang.NegativeArraySizeException␤ at org.perl6.nqp.sixmodel.reprs.VMArrayInstance.set_size_internal(VMArrayInstance.java:83)␤ at org.perl6.nqp.sixmodel.reprs.VMArrayInstance.bind_pos_boxed(VMArrayInstance.java:102)␤ at
..org.perl6.nqp.runtime.Ops.bindpos(Ops.java:1541)…
diakopter haha
uints anyone? 02:33
geekosaur I don't think java has unsigned types?
02:33 davidagain joined
diakopter geekosaur: you're right 02:34
timotimo hehe whoops
lue Apparently the solution is "use a bigger integer type" 02:35
diakopter nqp-jvm: my $a := []; $a[0.Int] := 1; say('alive'); 02:36
p6eval nqp-jvm: OUTPUT«Invocant must be an object␤ in <anon>␤ in process_args␤ in <anon>␤ in <anon>␤ in compile_op␤ in as_jast␤ in as_jast␤ in <anon>␤ in <anon>␤ in <anon>␤ in compile_op␤ in as_jast␤ in as_jast␤ in as_jast_clear_bindval␤ in compile_var␤ in as_jast␤ in as_jast␤…
diakopter nqp: my $a := []; $a[0.Int] := 1; say('alive');
p6eval nqp: OUTPUT«Method 'Int' not found for invocant of class 'Integer'␤current instr.: '' pc 62 ((file unknown):44) (/tmp/kMM1zv4CQx:1)␤»
diakopter nqp-jvm compiler bug
nqp-jvm: my $a := []; $a[9**9**2] := 1; say(nqp::elems($a)) 02:39
p6eval nqp-jvm: OUTPUT«java.lang.NegativeArraySizeException␤ at org.perl6.nqp.sixmodel.reprs.VMArrayInstance.set_size_internal(VMArrayInstance.java:83)␤ at org.perl6.nqp.sixmodel.reprs.VMArrayInstance.bind_pos_boxed(VMArrayInstance.java:102)␤ at
..org.perl6.nqp.runtime.Ops.bindpos(Ops.java:1541)… 02:40
diakopter nqp-jvm: my $a := []; $a[9**9**9] := 1; say(nqp::elems($a))
p6eval nqp-jvm: OUTPUT«VMArray: Can't resize to negative elements␤ in <anon>␤ in <anon>␤ in eval␤ in evalfiles␤ in command_eval␤ in command_line␤ in MAIN␤ in <anon>␤ in <anon>»
diakopter O_O
diakopter clueless, yet again
pmichaud TimToady: submitted talk... will keep you posted on developments. 02:41
I'm afk again for a while.
a reminder that we'd really like to see more p6 talks at yapc... please submit if you haven't already :) 02:42
diakopter nqp: my $a := []; $a[FFFFFFFFFFFFFFFFFFF] := 1; say(nqp::elems($a))
p6eval nqp: OUTPUT«1␤»
lue I found that JDK 8 is going to have unsigned int arithmetic (but not actual separate types), and there's also this implementation of unsigned ints: github.com/jOOQ/jOOU (in case an nqp-jvm dev is interested) 02:44
diakopter nqp: say($_[FFFFFFFFFFFFFFFFFFFFFFFFFFF]) 02:45
p6eval nqp: OUTPUT«Null PMC access in get_pmc_keyed_int()␤current instr.: '' pc 108 ((file unknown):166648018) (/tmp/Uko6rmQ45u:1)␤»
diakopter a roll is beneath me
nqp: say($_[0])
p6eval nqp: OUTPUT«Null PMC access in get_pmc_keyed_int()␤current instr.: '' pc 52 ((file unknown):43) (/tmp/7eRK01Rl6r:1)␤»
diakopter nqp: say($_.WHAT) 02:46
p6eval nqp: OUTPUT«Can only use get_what on a SixModelObject␤current instr.: '' pc 43 ((file unknown):39) (/tmp/xdCEgNK6OP:1)␤»
diakopter oh.
whatevs...
nqp-jvm: say($_.WHAT) 02:47
p6eval nqp-jvm: OUTPUT«Lexical '$_' not found␤ in <anon>␤ in <anon>␤ in eval␤ in evalfiles␤ in command_eval␤ in command_line␤ in MAIN␤ in <anon>␤ in <anon>»
diakopter \o/ <-- that's exasperation, not celebration, in this case
nqp-jvm: say(_.WHAT) 02:48
p6eval nqp-jvm: OUTPUT«␤»
diakopter wat.
nqp: say(_.WHAT)
p6eval nqp: OUTPUT«␤»
diakopter well, at least they're consistent
02:59 vaelxon joined 03:00 sqirrel__ joined 03:04 sqirrel_ left, rindolf joined 03:05 orafu left, OuLouFu joined, OuLouFu is now known as orafu 03:13 araujo left 03:25 census joined
diakopter bought my ticket to yapc;;na: submitted a talklet 03:27
census diakopter++
raiph diakopter++ # that's some "talklet" ;) (i'm subscribed to yapcna planning list so saw your submission) 03:29
colomon needs to accomplish a lot of contract work in the next few weeks to have a hope of affording YAPC::NA. 03:34
sorear how soon do I need to buy a ticket? i'm sure I can afford it but I am in a huuuge time crunch right now 03:41
colomon has no idea.
diakopter the current "early bird" price is $80. dunno when it rises
flights from SFO are pretty cheap right now, it seems 03:42
sorear i'll assume that was directed at me
SAN-SFO-Austin is a fairly substantial detour 03:43
oh wait, you live around SFO yourself
diakopter yes and no; I wasn't suggesting you go through SFO :P
03:56 xinming_ joined 03:59 xinming left 04:00 am0c left
census diakopter: SFO is the best, isn't it? :) 04:01
diakopter eh... 04:02
#@(@*@%%^$%*(#@(@$*%(@#* Facebook, Google, and Apple have driven up rents $1000/mo in the past 4 years
...not to mention buying up all the commercial real estate in that region; grrr 04:03
sorear diakopter: are you describing SFO or San Francisco?
diakopter oh. heh, oops.
census I did not say bay area! I said SFO airport!! 04:04
diakopter I guess I assumed census was talking about the area 'cuz I couldn't imagine anyone saying the airport is the best
census i love the international terminal 04:05
don't you?
diakopter oh; that's swank, yes
census sfo >> oakland!
geez
but are you telling me that you are a bay area resident who does not think that the bay area is the best place in the entire world??
diakopter uhm. 04:06
:) I grew up in Houston 04:07
sorear haven't been to SFO. int'l airports, by how pleasant my terminal stays were: LUX > FRA > SAN > ORD > LHR > LAX > TXL 04:09
04:10 census left
diakopter tip: if you're driving to SFO, always enter through the international terminal then walk to whatever terminal you need, 'cuz the security line there is always nonexistent compared to hours everywhere else 04:10
04:13 PacoAir left 04:16 census joined 04:24 _jaldhar joined 04:25 _jaldhar is now known as jaldhar 04:29 stevan joined
sorear just checked email. Um. 04:30
04:31 stevan_ left
diakopter census: are you in the bay area too? 04:33
census NO
diakopter ah, New Orleans
census that is not what i meant
i meant capital n o
diakopter I know :P
census as in no way
i just thought everybody in the world loved bay area 04:36
sorear where are you? 04:38
census working on a perl script :) 04:42
04:45 preflex_ joined 04:47 preflex left, preflex_ is now known as preflex 04:52 vaelxon left 05:16 ggoebel_ left, ggoebel_ joined 05:17 kivutar joined 05:48 isBEKaml joined
isBEKaml OHHAI, #perl6! 05:50
05:51 mavcunha left 05:52 renormalist joined
diakopter isBEKaml: 'sup 05:52
05:53 kaare__ joined 05:54 mavcunha joined
isBEKaml diakopter: Currently wrestling with an R* build for slackware for lack of proper destdir support. :) 05:54
what I have works, but module installs are F'ed up. :/ 05:56
diakopter via panda?
isBEKaml no, via makefile after rakudo-install step.
diakopter or the ones bundled?
isBEKaml (the ones bundled) - library locations are hardcoded into parrot-config which is looked up by modules-install(IIUC) 05:57
so, I'm thinking of fiddling with configure step or break the installs into two steps, one for the compiler and another for modules. 05:58
06:00 skids left
isBEKaml Okay, the installs themselves are fine, it's @*INC :| 06:09
sprunge.us/gDZV
the /tmp/package-rakudo-star is the sandbox folder used for building the R* slackware package. 06:11
Output from parrot_config --dump: sprunge.us/jERb 06:15
library install locations are all wrong. :/
/tmp/SBo/ should never be seen. I'm not sure how to proceed from here. 06:16
tadzik, moritz: please backlog. If you know of a solution, please let me know. 06:19
diakopter is it correct that $(PERL) appears in tht output? 06:20
isBEKaml diakopter: I see the same on cygwin install too. 06:21
Here's my script so far: raw.github.com/svatsan/RakudoStar-...SlackBuild 06:27
06:27 isomorphisms left
diakopter anyone around here a node user? 06:29
sorear meee 06:32
hi isBEKaml
isBEKaml hello, sorear
diakopter sorear: have you looked at the new Streams api in 0.10?
sorear no. 06:37
2/3 of our code is streams definitions, so i'm going to conjecture that we'll be on 0.9 forever 06:38
diakopter the announcement blog entry talks it up quite a lot; sounds impressively improved
it has a backcompat shim
diakopter stumbles across Microsoft's node.js hosting cloud thing; the pricing is nothing short of absurd 06:46
07:06 census left 07:14 spider-mario joined 07:30 rindolf left, rindolf joined 07:47 rindolf left, araujo joined, araujo left, araujo joined, SamuraiJack joined, rindolf joined 07:52 mberends joined, SamuraiJack_ joined 07:53 SamuraiJack left 07:59 rindolf left, rindolf joined 08:09 FROGGS joined
isBEKaml Anyone here a list admin on p6c? 08:43
08:47 xinming_ left 08:49 xinming joined 08:52 je joined 08:53 je left 08:59 isBEKaml left 09:01 xinming left 09:05 xinming joined
jnthn morning o/ 09:19
FROGGS morning jnthn
jnthn FROGGS: Your NFA usage causes the NFA evaluator to buffer overflow. ;) I fixed it yesterday :) 09:20
*caused
FROGGS ohh, wow
thanks, will pull in a sec
jnthn FROGGS: Surprised you didn't see heap corruption though... :)
It expected that [0] of @!states had as many elements as you have fates. 09:21
FROGGS jnthn: I dont the a commit in nom
jnthn FROGGS: The NFA evaluator is in the NQP reop
FROGGS ahh, I see
jnthn Anyway, nothing needs doing, unless you want to review it. :) 09:22
Mostly noting it in case you'd been trying to play hunt the SEGV
FROGGS surely I wanna see what you did :o)
sorear morning jnthn 09:25
jnthn ohhai 09:27
09:31 isBEKaml joined
diakopter jnthn: did you see the segv in the backlog? 09:31
jnthn > my $a = []; $a[240731133] = 1; 09:32
1
Is p6eval on a 32-bit machine?
diakopter think so
jnthn heh, the process where I did that has now allocated 2 gigs :) 09:35
.oO( did it quickly at least )
 09:36
FROGGS jnthn: btw, that are my local changes gist.github.com/FROGGS/d2c01ff6ada9efe61e07 (still need to switch to Perl6::P6Regex, and have a nicer way for :i) 09:37
jnthn: you said I need something like Compiler.eval but jump into regexes instead of comp_unit?
jnthn I guess what I'm curious about is exactly what the language is meant to be 09:39
r: my $x = 'abc { say 42 } def'; say 'abcdef' ~~ /<$x>/ 09:40
p6eval rakudo 1c9939: OUTPUT«42␤｢abcdef｣␤␤»
jnthn I wonder if the spec agrees that should work
FROGGS :/ 09:41
jnthn If so then yeah, what we're doing is really parsing Perl 6, just starting out in the regex slang. But in that case we still need all the usual setup in place to handle nested blocks. 09:42
I don't immediately find and S05 ruling on that 09:43
*an
bbi10 09:44
09:45 Liz_ left 09:46 Liz_ joined, Liz_ is now known as lizmat 09:47 lizmat left, spider-mario left
dalek nda/p6prove: e9d6331 | tadzik++ | / (7 files):
First shot at own-dogfooding prove
 09:49
tadzik testing welcome 09:50
isBEKaml hello, tadzik
tadzik hello isBEKaml
isBEKaml did you see the backlog about issues in packaging R* (with modules)? 09:51
tadzik yeah. I don't know where did this sbo come from
are you using some external installer, like your linux distro package manager?
isBEKaml This is for building R* packages on Slackware. 09:52
tadzik is it possible that this tool invents this path? 09:53
isBEKaml It goes like this: Slackware uses makepkg for generating packages - simply copying over all files from a predefined sandbox location(or folder), using relative path mechanisms to compress all files into a final .txz or .tgz package.
My problem is, the library installation paths are all absolute paths - they all point to temporary sandbox locations, which is wrong. 09:54
I got all of that info from parrot_config --dump
09:55 davidagain left
isBEKaml o.O(If I'm incoherent, this is not the first time :| ) 09:56
tadzik yeah, that's one of the problems; the installation is not relocatable 09:58
so you can't install it in one location and then move it elsewhere
unless you're in fakeroot, maybe
isBEKaml tadzik: this explains the build and install process: www.slackwiki.com/Building_A_Package 09:59
tadzik: IOW, doesn't respect --prefix and --destdir? :) 10:01
tadzik it's possible
isBEKaml It's all hardcoded in parrot_config :( 10:08
sorear isBEKaml: are you still looking for a mailing list admin? 10:11
isBEKaml sorear: no, it's handled. 10:12
/quit 10:25
10:25 isBEKaml left
masak good noon, #perl6 10:52
FROGGS hi masak
jnthn o/ masak 10:54
r: (1..16).grep({last if $_ % 5 == 0; $_ % 2 == 0}); 10:55
p6eval rakudo 1c9939: ( no output )
jnthn Does anybody get a segfault at exit running the above?
10:55 nyuszika7h left
jnthn is guessing it's just in md-nqp that it's happening... 10:55
masak sorry for spamming RT with private mail. :( I was tired, and the brain was apparently running on autopilot, picking the most common recipient I type into Gmail. 10:56
FROGGS jnthn: no
jnthn OK
Hmmm.
.oO( an unexpected downside of being Rakudo's leading bug filer... )
 10:57
FROGGS masak: is it something interesting that I have to read? :P 10:59
masak no, it's just an email that should have gone to my wife, not to RT. :( 11:01
please ignore.
FROGGS will do
11:05 nyuszika7h joined
sorear good morning masak. 11:06
masak: diakopter was very proactive and has already gotten it purged from nntp.perl.org and rt.perl.org
I however received a prompt copy, and not being in the best mental state myself spent a not insignificant amount of time trying to interpret the message as an allegory for some Rakudo problem 11:09
11:18 Chillance joined 11:20 FROGGS left
jnthn accidentally fixes a Rakudo bug while trying to hunt down the exit segfault... 11:22
11:24 sqirrel__ left, kivutar left 11:27 kivutar joined
sorear sleep& 11:31
11:39 sjn joined 11:44 kivutar left 11:45 kivutar joined
jnthn walk & 11:50
mberends transatlantic sleepwalk! 11:54
11:56 kurahaupo joined 11:59 kivutar left 12:02 kurahaupo left 12:03 kivutar joined
hoelzro ahoy #perl6 fork 12:04
*folk
so I'm trying to write a Perl 6 POSIX module, and I'm not sure how to call my module's functions 12:05
I have module POSIX { sub getuid() is native { ... } } in POSIX.pm
and if I add is export, I can call getuid() 12:06
but POSIX::getuid() doesn't work. Did the syntax for calling subs in modules change?
colomon I can verify that behavior. but I don't think I ever tried to use a full path for an exported sub before. hmm... 12:18
12:21 not_gerd joined
not_gerd o/ 12:21
not_gerd hands hoelzro an 'our'
not_gerd leaves again
12:21 not_gerd left
moritz what not_gerd said. Subs are lexical by default in Perl 6. 12:21
hoelzro ahhhhh 12:22
thank you!
can there be an analysis tool that detects when a my sub is never called?
also, is there a metamethod for iterating over a module's contents? like $obj.^methods? 12:23
I looked in ModuleHOW, but to no avail
jnthn ModuleName.WHO gets you a Stash 12:27
hoelzro ah ha
jnthn ModuleName::foo() is really ModuleName::<&foo>() which is really ModuleName.WHO<&foo> 12:28
hoelzro so let's say I want to throw exceptions on failure for my POSIX calls. Should I call my exception class X::POSIX, POSIX::X::Something, ?
jnthn X:: first seems to be the trend, afaict
hoelzro ok 12:29
hoelzro has working fork o/ 12:33
jnthn fork yeah! 12:34
hoelzro what would be "Perl 6"-y? returning an error value, or throwing an exception on failure? 12:35
jnthn Using fail(...) 12:36
hoelzro so much to re-learn =( 12:37
12:38 SamuraiJack_ left, SamuraiJack_ joined 12:44 snuffles left, snuffles joined 12:45 snuffles left 12:46 snuffles joined 13:02 PacoAir joined
masak sorear: hm, I didn't intent it as an allegory. but when you put it like that, that wouldn't have been completely unlike me. 13:10
grondilu rn: my (\k, \n) = 1, 2; say n-k; 13:12
p6eval niecza v24-34-g4f8f894: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'n-k' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1443 (die @ 5) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_unit @ 37) ␤ at /home…
..rakudo 1c9939: OUTPUT«===SORRY!===␤Undeclared routine:␤ n-k used at line 1␤␤»
13:12 domidumont joined
grondilu rn: my (\k, \n) = 1, 2; say n - k; 13:12
p6eval niecza v24-34-g4f8f894: OUTPUT«1␤»
..rakudo 1c9939: OUTPUT«===SORRY!===␤Error while compiling block : Error while compiling op call: Error while compiling block : Error while compiling op call: Error while compiling op call: Null PMC access in find_method('handle')␤»
jnthn RT'd already
I don't actually know how to fix that 13:13
Because I'd understood my \foo = blah; to be an initilization/declaration
And you bind the name directly
There's no container
And we compile ($foo, $bar) = ... down to normal list assignment iirc. 13:14
I was about to catch/forbid it, and then I saw Niecza actually makes it work. 13:15
n: my ($a, $b) = (my (\k, \n) = 1, 2);
p6eval niecza v24-34-g4f8f894: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/iDbVY6YNXH line 1:␤------> my ⏏($a, $b) = (my (\k, \n) = 1, 2);␤ $b is declared but not used at /tmp/iDbVY6YNXH line 1:␤------> my ⏏($a, $b) = (my (\k, \n) = 1…
jnthn n: my ($a, $b) = (my (\k, \n) = 1, 2); say k; $a = 42; say k; 13:16
p6eval niecza v24-34-g4f8f894: OUTPUT«Potential difficulties:␤ $b is declared but not used at /tmp/RC6hMHbIPE line 1:␤------> my ⏏($a, $b) = (my (\k, \n) = 1, 2); say k; ␤␤1␤1␤»
jnthn n: my $a := (my (\k, \n) = 1, 2)[0]; say k; $a = 42; say k;
p6eval niecza v24-34-g4f8f894: OUTPUT«1␤42␤»
jnthn ...oops. :) 13:17
n: my $a := (my (\k, \n) = 1, 2)[0]; say k; k = 42; say k;
p6eval niecza v24-34-g4f8f894: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/BGV0866aOw line 1:␤------> my ⏏$a := (my (\k, \n) = 1, 2)[0]; say k; k ␤␤1␤42␤»
jnthn needs to re-read the spec...
hm, hm, it doesn't say in the bit on it in S06... 13:19
Am I justly wildly misremembering that it has SSA-ish semantics?
masak .oO( wildly disremembering ) 13:21
jnthn: well, it does in the sense of "can only assign once".
jnthn masak: So it's nieczabug that I was able to assign to it a second time? 13:22
masak: Do you have a spec ref that actually says that, 'cus it's my recollection but I can't find the spec actually saying it...
masak hold on, lemmecheck. 13:26
masak .oO( be of some use for once )
:P
me, that is. :)
jnthn I had a panic at systemet today 13:30
uw, ww 13:31
masak you're just doing that to be nice :P
jnthn *uh. :)
masak S06:1250: In any case, it's a raw reference to whatever was passed, with no commitment
to structure or mutability.
jnthn No, but anyway, I went to buy stout and it was all gone.
masak so I guess \ vars aren't immutable.
jnthn But turns out they'd just moved it :)
masak oh phew :)
jnthn I guess it is on-topic here given I turn beer into (Perl 6 implementation) code :P 13:32
masak .oO( "not one of my stoutest moments" )
guess we'll have to explain "Systemet" to non-Swedes, too.
see, we lock our alcohol into a special place.
jnthn It's this place where you can buy beer at times the government see fit. 13:33
To be fair, it has a good selection.
masak yeah, it's not that bad actually. 13:34
and what they don't have, you can usually order.
jnthn I've lived in places where you could buy beer at any time of day, but if it's watery lager then, well, it's not quite such a win. :)
jnthn crosses his fingers and hopes he's about to find he has killed off the exit segfaults
masak except for the Westvleteren, which listerally sold out in *7 seconds*!
jnthn They had a HUGE yeti there today...aged in barrels. It was tempting but it cost 90 SEK for a pint bottle! 13:35
masak I'd have bought that one :)
jnthn I'm still thinking I shoulda 13:36
Ah well, I don't think it'll disappear right away.
masak .oO( Yetis tend to stick around... oh wait )
dalek p: 4fd9292 | jnthn++ | src/ (2 files):
Change the way we do invokewithcapture.

This avoids a bunch of nested runloops and all the problems that come with them. Re-working this, once a tweak is done in Rakudo, gets rid of the couple of exit-time SEGVs in the md-nqp branch.		 13:38
kudo/md-nqp: ba99183 | jnthn++ | src/core/MapIter.pm:
Fix LEAVE not firing on next/last in MapIter.

Done as an attempted/failed fix for something else.		 13:39
kudo/md-nqp: 706b65e | jnthn++ | src/binder/bind.c:
Fixes to work with the invokewithcapture changes.

With this, there's just one spectest (not file, just test) that seems to be broken in this branch over nom.
jnthn r: multi foo() { LEAVE { say 'bye' }; say 'x' }; foo() 13:40
p6eval rakudo 1c9939: OUTPUT«x␤»
jnthn That's fixed in the branch too.
dalek kudo/md-nqp: 103481f | (Timo Paulssen)++ | src/core/Exception.pm:
raise X::Method::NotFound even for '' methods.
 13:41
kudo/md-nqp: 1c99395 | jonathan++ | src/core/Exception.pm:
Merge pull request #117 from timo/typed_exc_empty_method

raise X::Method::NotFound even for '' methods.
kudo/md-nqp: 539d953 | jnthn++ | src/core/Exception.pm:
Merge branch 'nom' into md-nqp
jnthn
.oO( Good teamwork by jonathan++ and jnthn++! )
 13:42
Left to do: port the compile time multi-dispatch analysis to NQP, rip out remaining C-based multi-dispatch code, hunt the one remaining, evil, spectest regression. 13:44
Also, while you can't quite do anything useful with this yet, it's still funny: gist.github.com/jnthn/5176489 13:52
arnsholt That's fantastic! =D 13:54
13:55 LlamaRider joined
masak jnthn: you're our best dispatchee! :D 14:03
14:06 Heather joined
jnthn ;) 14:06
Heather hi
14:06 Chillance_ joined
jnthn The missing thing to make it useful is a way to get ahold of the capture and so forth. 14:06
Heather how to pass arguments to run ?
r: run <ls .> 14:07
p6eval rakudo 1c9939: OUTPUT«run is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting:2␤ in sub run at src/RESTRICTED.setting:8␤ in block at /tmp/VQXXo_nCQI:1␤␤»
jnthn run('notepad', 'README')
Heather ah it is )
< > works fine here 14:08
jnthn That did the Right Thing for me here
As right as using Notepad ever is, anyway ;)
14:10 Chillance left
Heather I forget how to write not_ok in test 14:10
jnthn nok I think. 14:11
r: use Test; nok(1)
p6eval rakudo 1c9939: OUTPUT«not ok 1 - ␤»
jnthn yeah
Heather ok 14:16
jnthn
.oO( was that a subtle pun... :) )
 14:18
14:23 Psyche^ joined 14:26 Patterner left, Psyche^ is now known as Patterner
Heather tadzik: added simple test github.com/Heather/ignore/blob/mas...1-ignore.t :) 14:30
LlamaRider r: class A{}; class B { use A; }; 14:36
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Could not find A in any of: /home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/site/lib, /home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/vendor/lib, /home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/lib, /home/p6eva…
LlamaRider Is that intended ? Is there a "one class per file" policy?
jnthn use always means "load a module" 14:37
You can have as many classes as you want in it.
It's up to you if the path/filename of the module (as in, file of code) and the classes etc declared inside of that file are in any way related. 14:38
So use A really means "look for an A.pm, load it, and import anything that it exports by default into the current lexical scope" 14:39
14:40 SamuraiJack__ joined
hoelzro I'm continuing work on my POSIX module; how can I set it up so that nothing is imported with 'use POSIX', but you can import functions like so: 'use POSIX <fork>' 14:41
14:41 kivutar left
hoelzro I currently have our sub fork() returns Int is export(:ALL) { ... } 14:41
jnthn You can make your own tags with things like is export(:mything)
LlamaRider jnthn++ ok, I see. Will need to refactor things more sanely then. And I guess putting class A::B in A/B.pm and having a one class per file policy would induce sanity. 14:42
hoelzro jnthn: but you can't import them one-by-one? 14:43
14:43 SamuraiJack_ left
jnthn hoelzro: Hm, good question. use Foo <bar> passes it as a positional argument to some custom EXPORT sub 14:44
hoelzro is 'use Foo <bar>' not Perl 6-y?
14:44 SamuraiJack__ left 14:45 SamuraiJack__ joined
jnthn So far as I can see from what's implemented, it isn't doing anything that lets you list the symbols you want, at least... 14:46
If it were you'd probably have to say &bar anyway
hoelzro I see 14:47
jnthn oh, hmm
hoelzro so for now I should implement EXPORT myself? 14:48
hoelzro will return later 14:50
jnthn wonders if moritz++ knows a bit more about how this works/should work :)
Heather r: my $o = [2, 2]; $o >> { say . } 14:54
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Unsupported use of >> to do right shift; in Perl 6 please use +> or ~>␤at /tmp/2YmdznwQQk:1␤------> my $o = [2, 2]; $o >>⏏ { say . }␤»
Heather r: my $o = [2, 2]; $o>> { say . } 14:55
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Unsupported use of >> to do right shift; in Perl 6 please use +> or ~>␤at /tmp/URs35nWPlh:1␤------> my $o = [2, 2]; $o>>⏏ { say . }␤ expecting any of:␤ postfix␤»
Heather r: my $o = [2, 2]; $o>>{ say . }
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at /tmp/GUGa802n9J:1␤------> my $o = [2, 2]; $o>>{ say . ⏏}␤»
jnthn r: my $o = [2, 2]; $o>>.say
p6eval rakudo 1c9939: OUTPUT«2␤2␤»
Heather r: my $o = [2, 2]; $o>>.=1
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Missing << or >>␤at /tmp/LXKYJw9k7K:1␤------> my $o = [2, 2]; $o>>.=⏏1␤ expecting any of:␤ dotty method or postfix␤» 14:56
Heather jnthn: thanks how to override them?
jnthn What exactly does "override" mean?
Heather jnthn: make [1, 1]
jnthn Oh, assign to all of the things in the array? 14:58
= is an infix so it needs to be inside a hyper 14:59
r: my $o = [2,2]; $o >>=>> 1;
p6eval rakudo 1c9939: OUTPUT«===SORRY!===␤Missing << or >>␤at /tmp/HwiOta0Xfq:1␤------> my $o = [2,2]; $o >>=>⏏> 1;␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤»
jnthn hmm
r: my $o = [2,2]; $o >>[=]>> 1;
p6eval rakudo 1c9939: ( no output )
jnthn std: my $o = [2,2]; $o >>=>> 1;
p6eval std 86b102f: OUTPUT«===SORRY!===␤Missing << or >> at /tmp/GpOeO2CRi0 line 1:␤------> my $o = [2,2]; $o >>=>⏏> 1;␤ expecting infix_circumfix_meta_operator␤Parse failed␤FAILED 00:00 45m␤»
jnthn r: my $o = [2,2]; $o >>[=]>> 1; say $o 15:00
p6eval rakudo 1c9939: OUTPUT«1 1␤»
jnthn The [...] go around the = to disambiguate here
I think it's parsing => inside of there
arnsholt That'd be the longest token, I guess 15:01
jnthn yeah
arnsholt Once again clear and unambiguous rules fall flat in the face of something that's obvious to people =)
masak LlamaRider: 'use A' goes looking for an A.pm somewhere 15:02
oops, replied to backscroll.
LlamaRider masak: Thanks anyway, I'm writing this up in my porting repo now. To me it is a P6 gotcha. 15:04
masak p5 works the same. 15:05
LlamaRider then I got used to a P5 bug and thought it was a feature.
masak files are orthogonal to packages. and 'use' loads files. 15:06
15:06 census joined
LlamaRider the example I made comes from XML::LibXML and I've written code like that myself, where I never cared about use-ing something multiple times. Maybe P5 just ignores a file if already loaded? 15:07
or alternatively the code is a bug waiting to happen
census llamarider++
masak yes, p5 (and p6) won't care about an already loaded file. 15:08
LlamaRider well p6 does, it gave me a circular load error
masak also, 'if $cond { use SomeModule; }' will always load SomeModule.pm -- if you're unsure why, consider when the module is loaded, and when the if statement runs :)
LlamaRider: there should never be any circular load errors. 15:09
that sounds like an error in the implementation.
jnthn Huh.
Heather ingy: jnthn: oh, thank you )
jnthn If you use A; in a module B and you use B inside of a module A, that won't work out.
We don't reload from disk already loaded things. 15:10
geekosaur perl5 also has some interesting behaviors with circular imports, although it doesn't throw errors (warnings though...)
jnthn Well, in Perl 6 we simply can't do it
LlamaRider ===SORRY!===
Circular module loading detected involving module 'XML::LibXML'
jnthn Right.
Maybe the message should be a bit more like "you can't do this"
Rather than "oh, I noticed that..." :) 15:11
LlamaRider p5eval: package A; 1; package B; use A; 1; 15:14
p5eval LlamaRider: ERROR: Can't locate A.pm in @INC (@INC contains: /lib) at (eval 7) line 1.BEGIN failed--compilation aborted at (eval 7) line 1.
geekosaur right, use doesn't work that way in p5
LlamaRider but but... 15:15
$ perl -e 'package A; 1; package B; use A; 1; print "Works!\n"; '
Works!
geekosaur what I got with my inadvertent circularity is it loaded them, but the second load got a bunch of "Sub ... already defined"
LlamaRider it works on my machine... somehow
geekosaur different version? I was using whatever perl is on centos6.3
I think that's 5.10.mumble
LlamaRider I have 5.14 here 15:16
geekosaur and I know for certain it didn't work in 5.8.8 15:17
so apparently that got fixed at some point
jnthn LlamaRider: Do you happen to have some kind of A.pm hanging around?
LlamaRider hides 15:18
I did
now it breaks again. Sorry
LlamaRider silently goes back to porting
jnthn np ) 15:19
masak jnthn: is there any reason we don't just silently ignore module circularities in Perl 6, like we do in Perl 5? 15:21
geekosaur p5 is not silent (you get a bunch of sub redefined warnings), and you can get weird behavior if any of the modules do tricky stuff in BEGIN blocks 15:22
jnthn masak: Because you're almost certainly doomed to parse fail as soon as you mention any symbol form the other module, 'cus we didn't parse it yet. 15:23
nwc10 jnthn: selftest passes
15:23 am0c joined
masak jnthn: oh, and Perl 5 relies more heavily on strings there. I see. 15:23
LlamaRider wows at the P5 idioms in XML::LibXML 15:28
15:28 census left 15:33 census joined 15:38 timotimo joined 15:39 xenoterracide joined 15:42 pmurias joined
jnthn timotimo: ping 15:51
dalek kudo/md-nqp: fbf17f1 | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Port compile-time dispatch analysis to NQP.
 15:52
kudo/md-nqp: 8d0bb45 | jnthn++ | src/Perl6/Optimizer.pm:
Get optimizer using NQP-based dispatch analysis.
kudo/md-nqp: fdc0c2f | jnthn++ | / (5 files):
Finish eliminating C-based multi-dispatcher.
16:13 wk_ left 16:17 wk_ joined
nwc10 does this really work? 16:22
p5eval: CORE::dump
p5eval nwc10: No output
16:22 cogno joined
nwc10 is it patched for CVE-2013-1667 ? 16:23
masak works locally.
diakopter au++ obra++ ask++ 16:24
TimToady masak: when you use a module, you're assumed to be mutating your language. a circular use structure would then imply that you're new language is determined by a module that you didn't really know the language of when you used it. hence it is considered undefined, and pink zebras will fly out of your nose 16:25
nwc10 is wondering just how pwnable it is
TimToady s/you're/your/
.oO(what's black and white and pink all over?)
 16:26
geekosaur um, isn't that expected?
("dump" is a p5 builtin which used to be used with undump to create "preloaded perls") 16:27
jnthn read that as "pink zebras will fly out of your mom" :P
geekosaur as such, if there is a bug, it's that (a) it still exists (b) someone, somewhere, is still using undump 16:28
pmurias TimToady: shouldn't we detect that instead of crazy thing happening? 16:29
16:29 cogno left
jnthn pmurias: Rakudo does already :) 16:29
Though they error may want some wordsmithing.
nwc10 p5eval: print unpack "P1", ("\xFF"x8) 16:30
p5eval nwc10: No output
16:31 cognominal joined
masak TimToady: does that mean that p5 was confused about the "a module mutates your language" bit? 16:32
s/was/is/
geekosaur p5 had no such concept
packages are packages, semantics kinda just grew around them without much regard for the notion of language 16:33
16:35 pmurias left
masak TimToady: when I re-read your reasoning, it seems to me that the rationale somehow misses the common case, namely "oh, the user wants this module included, but it's already included. ignore" 16:35
dalek kudo/md-nqp: c57f0d5 | jnthn++ | src/Perl6/World.pm:
Don't re-link already compiled blocks.

This loses lexical chain information they are carrying. A test for traits/wrapping/multi interaction (yes, we have one :-)) seems to have been somehow magically working before, for mysterious reasons. This change makes it work for a better understood reason, and with the NQP based multi-dispatcher. Fixes the remaining regression in this branch, and introduces no more spectest regressions.
masak TimToady: there might even be *legitimate* cases of mutual inclusion (and ignoring one of the inclusions to avoid a circularity): when you have different roots in the inclusion hierarchy, for example.
jnthn masak: Isn't the previous comment a contradiction? 16:37
16:37 mavcunha left
jnthn masak: It can't be "already included" if it has a circular dependency because we never got to the point of having "included" it 16:37
masak hm, it seems to me there are two different cases there. 16:38
jnthn The point we consider a module loaded is after we have parsed it. At that point, we enter its UNIT into the "stuff we loaded" table.
masak let's say A has a 'use B'. if we encounter 'use A' while before finishing processing B.pm, then that's probably an error. 16:39
of the kind TimToady describes.
jnthn Agree so far. 16:40
masak but we could also have a situation where 'use A' happens after we've finished processing B.pm.
I suppose that's not a problem at all, and that case gets ignored.
jnthn We don't reload things a second time if they're already loaded. But that's not circular, that's just multiple things having a "use" 16:41
Really it's the "need"'s work that is cached
It's the import part of the "use" that needs doing each time.
OK, my branch seems to be working nicely :) 16:42
16:43 cognominal left
masak hm. yes. maybe that *is* the best we can hope for. 16:43
TimToady circular needs are probably okay, since they don't mutate the language
masak "Perl 6: it mutates your language"
this is true on so many levels. :)
jnthn 26 files changed, 1158 insertions(+), 1669 deletions(-) 16:44
You can write a multi-dispatcher in less lines in NQP than in C. :)
colomon \o/
TimToady I should hope so
now make it run faster :)
jnthn "Don't be a load slower" has been some work :) 16:45
masak wow, cool. we should have this: metacpan.org/module/PerlIO::http :)
jnthn But yes, more optimization needed/to come :)
It's not far off the C version though, and it fixes some bugs.
nwc10 cool 16:46
jnthn And gets rid of a load of places we did nested runloops
dalek ast: bbe88fe | jnthn++ | integration/weird-errors.t:
Correct an expected exception type.

After timotimo++ fixed things to throw the correct one.		 16:52
16:53 cognominal joined 16:56 SamuraiJack__ left 16:58 SamuraiJack joined
colomon TimToady: have you taken a look at the spectest issues I found with niecza yesterday? 16:58
in particular, 16:59
rn: say ($_ * 2 if $_ ** 2 > 3 for 0 .. 5)
nwc10 panda requires prove at the moment? If so, how SMOP is it to re-implement prove in Perl 6? Seems like a useful task
p6eval niecza v24-34-g4f8f894: OUTPUT«() () 4 6 8 10␤»
..rakudo 1c9939: OUTPUT«4 6 8 10␤»
jnthn nwc10: tadzik++ has already started working on doing that :)
colomon tadzik++ indeed!
nwc10 excleent 17:00
tadzik++
dalek Heuristic branch merge: pushed 47 commits to rakudo/nom by jnthn 17:01
nwc10 is that the goalposts moving again? and roughly in the direction of Christmas?
jnthn nwc10: That's one big piece of Rakudo just becoming a huge amount easier to port. Plus fixing some bugs for a more immediate win :) 17:02
nwc10 yay \o/ 17:03
jnthn++
colomon btw, the STEP script is still running, with several STEP files reading in successfully after 10+ hours of processing. so it's way too damned slow, but given enough memory resources it seems pretty robust! 17:04
jnthn++
current STEP file has been processing for
r: 983/60 17:05
p6eval rakudo 1c9939: ( no output )
colomon r: say 983/60
p6eval rakudo 1c9939: OUTPUT«16.383333␤»
timotimo jnthn: pong
colomon sixteen hours, and is currently using 25 gigs of memory.
cognominal jnthn, can you blog your wonderful work? It is so nice to understand what you are up to.
jnthn timotimo: I was gonna ask about a test, but I've sorted it out.
timotimo: One of your improvements caused a test to fail, but it was the test to blame
timotimo: See my last roast commit to know which one :)
pmichaud good afternoon, #perl6 17:06
diakopter o/
timotimo ah, what a relief
masak pmichaud! \o/
colomon o/
jnthn o/ pmichaud 17:07
uh-oh...I merge a big branch and Pm shows up :)
masak nwc10: I think we should definitely try to implement 'prove' in Perl 6.
pmichaud jnthn: my appearance has nothing to do with the branch merge :) 17:08
jnthn pmichaud: phew ;)
nwc10 and once it works for tests one at a time, parallel tests...
jnthn pmichaud: It didn't do much. Just re-wrote the multi-dispatcher from C into NQP :)
nwc10 which will be "fun" on Win32 thanks to Microsoft's utter [censored] w.r.t. the behaviour of sockets on close() 17:09
pmichaud jnthn: yes, I've been following the commit logs :)
timotimo jnthn: the rewrite is complete?
jnthn timotimo: Yes, and merged 17:10
One more bonus: we now have typed exceptions for multi-dispatch failures.
timotimo that is very nice 17:12
what's the next target on the road to rakudo on the jvm?
cognominal why the C based dispatcher was an hindrance for rakudo on the jvm? 17:13
timotimo i guess it was very parrot-centered?
nwc10 the C based dispatcher was in C 17:14
jnthn cognominal: Because it'd have had to be ported to something that runs on the JVM, and then we'd have had to maintain two implementations of it into the future.
nwc10 NQP runs on the JVM
C doesn't
jnthn cognominal: And when one of those things can be NQP, which runs on both... :)
nwc10 "the C based <mumble> is in C" is the problem, for all values of <mumble>
timotimo oh, there's a perl mongers group in my home town 17:15
nwc10 where in the world are you (or them)?
timotimo karlsruhe, both me and them
nwc10 aha
timotimo in baden-württemberg of germany
jnthn Oh, I changed train in Karlsruhe twice last year
nwc10 it's on TGV Ost, isn't it? 17:16
timotimo i think a TGV goes by there, too
but i've never seen or rode a TGV
jnthn Yeah, I was heading to last year's FPW 17:17
timotimo that is the french perl workshop?
jnthn ja
timotimo i wonder if they would be welcoming someone who has only very basic experience with perl 6 and none with perl 5 17:18
i would certainly not enjoy bringing a 5 vs 6 flame war into a group of people
dalek ecza: 5c06e28 | larry++ | lib/CORE.setting:
make Empty gist to '' like normal ()
17:18 Heather left
cognominal the FPW is in Nancy this year, north east of France, like last year. journeesperl.fr/fpw2013/ Perl6 speakers are welcome 17:19
beside the usual Perl6 speakers, having lizmat and wendy was nice, last year 17:20
masak "north east of Paris"? 17:21
cognominal to restate it, fpw : June 14-15, 2013 in Nancy
masak++ # so true 17:22
masak aww, I can't on those dates. and neither can jnthn. :/
woulda been nice to show up. 17:23
nwc10 jnthn: all tests and all spectests other than t/spec/S26-documentation/09-configuration.t pass
the failures are all like this:
ok 1 - The object is-a 'Pod::Block::Code'
not ok 2 -
# got: '1'
# expected: 'B'
got '1' instead of the thing expected
(apparently it is now tea time) 17:24
jnthn nwc10: Yeah, that's not a regression due to the branch, it was an issue in nom too. I've currently compiling a patch...I think some of my earlier work busted it
masak: DDDx?
moritz lib/lib.pm6 has a custom EXPORT sub 17:25
masak jnthn: ayup.
\o/
jnthn Nice :) Though pity we miss FPW.
moritz: I don't think it actually installs symbols, though? :) 17:26
moritz: Or it wasn't obvious to me how it would...
Yay, I got a fix for that Pod test. :) 17:27
Maybe now Rakudo can haz clean roast runs again
masak \o/ 17:30
pmichaud I get a build failure in nom :-/ 17:31
jnthn Congrats! 17:32
What's it look like?
oh...argh
pmichaud it's been a while since I last built... starting with clean checkouts
(restarting)
dalek kudo/nom: 52f04a7 | jnthn++ | src/Perl6/Pod.pm:
Fix Pod configuration regression.
 17:33
kudo/nom: fd5d929 | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION.
jnthn pmichaud: I think I forgot to bump NQP_REVISION recently enough.
pmichaud looked like that might've been it
running 'make' now 17:40
grondilu rn: .say for 1 .. ^"foo".chars 17:41
p6eval rakudo 1c9939, niecza v24-34-g4f8f894: OUTPUT«1␤2␤3␤»
grondilu rn: .say for ^("foo".chars)
p6eval rakudo 1c9939, niecza v24-34-g4f8f894: OUTPUT«0␤1␤2␤»
grondilu rn: .say for ^"foo".chars
p6eval rakudo 1c9939, niecza v24-34-g4f8f894: OUTPUT«0␤1␤2␤»
timotimo ^"foo" is evaluated as ^+"foo"" no? 17:42
pmichaud rakudo: say ^"foo"
p6eval rakudo 1c9939: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: Mu *%_)␤␤ in method Real at src/gen/CORE.setting:873␤ in method new at src/gen/CORE.setting:5281␤ in sub prefix:<^> at src/gen/CORE.setting:5474␤ in block at /tmp/1AIqgfkj0c:1␤␤»
pmichaud no. 17:43
grondilu understood his mistake
I meant:
rn: .say for 1 ..^ "foo".chars
p6eval rakudo 1c9939, niecza v24-34-g4f8f894: OUTPUT«1␤2␤»
grondilu yep
better
timotimo oh, duh, that's obvious 17:44
grondilu though I'm not sure what 1 .. ^"foo".chars means and why it returns 1, 2, 3
pmichaud ^"foo".chars is a Range 17:45
grondilu rn: .say for 1 .. (^3)
p6eval rakudo 1c9939, niecza v24-34-g4f8f894: OUTPUT«1␤2␤3␤»
grondilu ^ guess that's it
pmichaud: yep, ^"foo".chars is a range, converted in integer by '..' 17:46
17:46 cognominal left
pmichaud istr that we were going to fail with Range arguments to infix:<..> 17:46
pmichaud checks the spec
timotimo jnthn: did you manage to find out why the nqp md was slower than the c md? 17:47
jnthn timotimo: It isn't by terribly much any more. 17:48
timotimo nice :)
pmichaud as a general rule of thumb I'd expect nqp <mumble> to be slower than C <mumble>
timotimo sure
jnthn Yeah
Well, there's "slower" and there's "omg we can no way merge this" slower :)
pmichaud right 17:49
timotimo can you decide at compile time if you go with C md or nqp md? or was C md removed altogether?
jnthn Removed
17:49 cognominal joined
timotimo hm, i can see why you would do that. 17:49
pmichaud yes, S03:3365 says it's illegal to use a Range as an endpoint of infix:<..>. I guess that needs a multisub entry... will work on it after I get Rakudo running :) 17:51
what's the standard process for identifying and/or creating an exception, ooc?
is there an article somewhere? 17:52
timotimo is ooc "out of curiosity"?
pmichaud I'm assuming that fail "string describing failure" is no longer acceptable :-)
ooc == "out of curiosity", yes
moritz pmichaud: you create one, document it in S32::Exceptions, test it (in case of doubt t/spec/S32-exceptions/misc.t)
jnthn missed two of those steps...d'oh :) 17:53
pmichaud moritz: any other guidelines or rules of thumb?
jnthn While we're on the topic, though...
moritz and if you are motivated, also document it in the perl6/doc repo
pmichaud: use common sense
and compile time exceptions should do X::Comp
jnthn rt.perl.org/rt3/Ticket/Display.html?id=114204 suggests that we should include info on what arguments you tried to pass in the case of a failed multi-dispatch. 17:54
Do we want that? 17:55
moritz +1
jnthn If so I can bundle the info into the exception objects and then it should be easy to stringify...
moritz aye
maybe jut the types, and names in case of named arguments
jnthn *nod* 17:56
pmichaud so, in the case of 1 .. ^10 what sort of exception would be expected? Message is something like "Ranges are not allowed as endpoints to &infix:<..>"
X::RangeOp::InvalidArg ? 17:57
moritz X::Range::InvalidArg I'd say
jnthn When the error is clearly related to the type it tends to be X::TypeName::Something
pmichaud oh, so Range.new(1, ^10) would be an error. okay. 17:58
jnthn X::Buf::AsStr for example
moritz pmichaud: yes, I don't see how Range.new(1, ^10) makes more sense than 1..(^10)
pmichaud okay, makes sense to me 17:59
jnthn On the FooOp pattern though, we do have a X::HyperOp::NonDWIM 18:00
moritz but we don't have a Hyper type
pmichaud right, I was looking for an op pattern.
jnthn aye :)
pmichaud I agree it makes sense to do it in Range here.
indeed, Range.new already has an unconverted fail "message" that I'll also take care of. 18:01
18:01 cognominal left
pmichaud r: say Range.new(*,*) 18:01
p6eval rakudo 1c9939: OUTPUT«*..* is not a valid range␤ in method gist at src/gen/CORE.setting:10555␤ in sub say at src/gen/CORE.setting:7614␤ in block at /tmp/fXUM4HaHTm:1␤␤»
pmichaud running spectest now of rakudo with new dispatch :) 18:02
nwc10 jnthn: works on my machine (spectests)
pmichaud no failures thus far... up to S04
jnthn yay :)
masak heh, why isn't *..* a valid range? 18:05
grondilu failed to solve yet another rosalind problem because perl6 is too slow
masak r: say 5 ~~ *..*
p6eval rakudo 1c9939: OUTPUT«False␤»
masak n: say 5 ~~ *..*
p6eval niecza v24-35-g5c06e28: OUTPUT«True␤»
masak hah! :D
moritz which one is right?
masak tentatively argues Niecza is right and *..* is valid 18:06
jnthn Wait, how come Range.new(*,*) doesn't work, but *..* does?
18:06 cognominal joined
pmichaud masak: *..* isn't valid because S03:3468 says it's Illegal. 18:06
jnthn rn: for *..* { .say }
masak pmichaud: fair enough :)
p6eval rakudo 1c9939: OUTPUT«*..* is not a valid range␤current instr.: 'throw' pc 334686 (src/gen/CORE.setting.pir:150131) (src/gen/CORE.setting:9331)␤called from Sub 'gist' pc 366646 (src/gen/CORE.setting.pir:162889) (src/gen/CORE.setting:10555)␤called from Sub 'say' pc 80552 (src/gen/CORE.se…
..niecza v24-35-g5c06e28: OUTPUT«(timeout)-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-I…
grondilu paste.siduction.org/20130316180629 << how could I improve the speed of this program?
pmichaud S03:3463 says it also :) 18:07
masak aye.
pmichaud grondilu: removing the 'return's would speed things up
moritz grondilu: 'return [+] gather' looks scarily slow 18:08
grondilu: instead declare a variable and use it to sum up stuff
masak rn: my $counter = 0; for *..0 { .say; last if $counter++ == 10 }
TimToady r: say -Inf .. Inf
grondilu ok
pmichaud return $rna eq any(<AU UA CG GC>) ?? 1 !! 0 is much faster as +($rna eq any(<AU UA CG GC>))
p6eval rakudo 1c9939: OUTPUT«(timeout)»
..niecza v24-35-g5c06e28: OUTPUT«-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤»
rakudo 1c9939: OUTPUT«-Inf..Inf␤»
pmichaud well, maybe not _much_ faster, but faster.
masak both Rakudo and Niecza seem wrong here. 18:09
TimToady -Inf .. Inf is acceptable
jnthn But useless for iterating :)
TimToady it's not for iterating, but then nothing with * on the left is either 18:10
it's for range matching
pmichaud jnthn: all spectests passed here
jnthn \o/
TimToady especially if something requires a Range, but you want to match everything, then -Inf .. Inf can be useful
masak rn: say 5 ~~ -Inf .. Inf
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«True␤»
masak rn: say 5 ~~ * .. Inf
18:11 cognominal left
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«True␤» 18:11
masak rn: say 5 ~~ * .. *
p6eval rakudo 1c9939: OUTPUT«False␤»
..niecza v24-35-g5c06e28: OUTPUT«True␤»
TimToady two bugs for the price of one 18:12
grondilu rn: say +("foo" eq any(<foo bar>));
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«any(1, 0)␤»
grondilu pmichaud: ^
pmichaud grondilu: oh
rn: say +?("foo" eq any(<foo bar>))
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«1␤»
TimToady rn: say +so "foo" eq any <foo bar> 18:13
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«1␤»
pmichaud rn: say True * False # curious
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«0␤»
18:14 Sifr joined
pmichaud rn: say True * True 18:14
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«1␤»
grondilu retries with suggested modification and full data
pmichaud might not need the + :-)
timotimo how hard would it be to optimise things like ([+] gather) or for ..Inf -> $x { ... } where a lazy list is generated but the earlier pieces of the list can be discarded immediately 18:15
is this something the programmer should be aware of and avoid actively, or should the compiler notice what's going on and cleverly deallocate the early parts of the list?
pmichaud rn: say True +^ False
p6eval rakudo 1c9939, niecza v24-35-g5c06e28: OUTPUT«1␤»
moritz timotimo: it's at least not trivial to optimize 18:16
pmichaud timotimo: "cleverly deallocate" is really hard at the moment.
I need to get back to my list updates :-/
timotimo i feared as much
pmichaud bbiab 18:17
dalek kudo/nom: ce914fc | jnthn++ | src/Perl6/ModuleLoader.pm:
Some pir:: => nqp:: in Perl6::ModuleLoader.
 18:18
jnthn ye shoppe & 18:19
nwc10 1 files changed, 5 insertions(+), 8 deletions(-)
not only pir:: => nqp::, but also net code removal
grondilu six minutes and still running :( 18:20
grondilu interrupts the program 18:21
moritz grondilu: you can substitute .match($regex, :g) with .comb($regex)
which returns Str, not Match objects, so uses less memory
grondilu ok 18:22
I doubt it will help much though, as this line only reads the input
anyone has a fast machine and would like to give it a go? 18:23
timotimo Q:PIR { $P0 = find_lex '$block'; $P1 = find_lex '$coro'; $P1($P0) }; - is this just basically "$coro($block)"?
grondilu Here is the code: paste.siduction.org/20130316182329 18:24
and here is the input: paste.siduction.org/20130316182413
18:25 LlamaRider left 18:30 cogno joined 18:32 FROGGS joined 18:35 cogno left
jnthn grondilu: Guess I can feed it through the profiler while I cook :) 18:36
timotimo grondilu: how does the "bag max :by(*.chars) thing work? i'm confused
grondilu timotimo: it checks that there are as many A as U and C as G. I'm currently trying to avoid doing this check 18:37
timotimo i don't even get how it parses
pmichaud timotimo: ( $coro($block) ) Yes, that's what it basically is, but it goes through the Parrot dispatcher and not the Rakudo one.
timotimo is :by a named parameter to the max function?
grondilu timotimo: yes
timotimo pmichaud: so i should replace it with the appropriate nqp ops? (i think if i ask one more question, you'll have done 100% of the work >_>) 18:38
grondilu rn: say max :by(*.chars), <foo bar longstring>;
p6eval rakudo fd5d92, niecza v24-35-g5c06e28: OUTPUT«longstring␤»
timotimo i see
pmichaud timotimo: if there are appropriate nqp ops, then yes. 18:39
note that $coro is probably not a Code object.
that's the part to be aware of.
timotimo at this point, the $coro is a Mu that was nqp::clone'd from nqp::getattr(&coro, Code, '$!do'), so it seems like it really is a Code object 18:40
oh, no, hold on
the second argument is probably the class from which the attribute comes, right? 18:41
timotimo doesn't know how to continue
pmichaud timotimo: yes, Core is the class defining the attrib ute 18:44
for any Code object, $!do is the low-level subroutine thingy to be invoked
I'm assuming you're working in GatherIter 18:46
if so, then $coro is the Parrot Coroutine PMC to be invoked for this particular GatherIter 18:47
timotimo i just quickly grepped for PIR and this was the first result ;)
18:49 MayDaniel joined
grondilu has translated in P5 and even then it is slow. The code is probably wrong. 18:59
grondilu realizes he forgot to translate a if case 19:05
19:07 wk_ left
arnsholt o/ 19:07
masak \o
19:16 cognominal joined
pmichaud okay, I'm having trouble understanding the dispatcher 19:16
or, at least, what I'm doing wrong, if anything
gist.github.com/pmichaud/5177871 # code added to src/core/Range.pm 19:17
this adds a .new candidate that throws an exception if the second argument is a Range 19:18
but when I run it, it continues to call the existing candidate
(the one without type constraints)
> say Range.new(0, ^10) 19:19
0..10
what am I overlooking?
timotimo r: my ($a, $b, $c, $d) = <a b c d>; ($a, $b, $c, $d) .= pick(4); say "$a, $b, $c, $d"; 19:22
p6eval rakudo ce914f: OUTPUT«c, d, b, a␤»
timotimo r: my ($a, $b, $c, $d) = <a b c d>; ($a, $b, $c, $d) .= pick(4); say "$a, $b, $c, $d";
p6eval rakudo ce914f: OUTPUT«d, a, c, b␤»
timotimo this amuses me
19:22 spider-mario joined
pmichaud ...and apparently it matters _where_ I put the method. 19:24
arnsholt jnthn: Does gist.github.com/arnsholt/5177904 look like a reasonable translation of bootType from NQP/JVM? 19:25
pmichaud If I put it first or second, then .new(0,^10) triggers the method. If it goes after method new(Real, Any, ...) then it never gets selected. 19:26
19:27 am0c left
pmichaud oh, I see why. I think. 19:28
.new(0,^10) chooses whichever comes first from new(Real,Any) and new(Any,Range)
jnthn Those would be tied 19:29
pmichaud right.
jnthn And then the presence of named args is enough to throw the binder into "first that binds" mode. 19:30
pmichaud okay, that's the piece I was missing then.
jnthn sorry, named params 19:31
mberends
.oO(named parens)
 19:33
19:34 kurahaupo joined
pmichaud so, what's the best approach here? create sufficient candidates to make sure there aren't any ties, or order the methods in terms of preference? 19:34
jnthn Unicode probably knows hundreds of different named kinds of parens... :)
jnthn checks what candidates we have already
arnsholt I was thinking that's what happens when you program too much in Lisp. Every paren gets a name ^_^
mst I shall call mine ... timmy! 19:35
this is your parenthesis ... there are many others like it, but this one is yours ...
19:35 quester joined
mberends like Nemo's 400 siblings 19:36
jnthn pmichaud: Alternatively, you could check $max ain't a Range in the BUILD submethod. 19:37
hoelzro so I'd like to define a custom EXPORT routine for my module. I thought that required me to do something like this: module POSIX { our sub EXPORT(*args) { ... } }, but that doesn't work
(that doesn't work -> "no EXPORT sub, but you provided positional argument in the 'use' statement")
jnthn pmichaud: Which covers all of the bases
hoelzro: I think it may be looked up in UNIT
pmichaud jnthn: yeah, I thought about that a bit also. 19:38
jnthn It kinda cross-cuts the other candidates.
pmichaud that seems a bit too low-level to me somehow, but I'll go ahead and do it that way for now.
hoelzro jnthn: tried that too; "Merging GLOBAL symbols failed: duplicate definition of symbol &EXPORT"
I'm guessing it doesn't play nice with 'is export'?
jnthn hoelzro: I don't think you're meant to export your EXPORT :) 19:39
pmichaud: I suspect putting the candidate after the first Any,Any one would do the trick.
hoelzro right, but I also have our sub getuid() returns Int is native is export
moritz hoelzro: note that sub EXPORT must be outside the module 19:40
hoelzro right, I'm doing that
pmichaud yeah, putting it after the first Any,Any seems to work, but also seems fragile.
is the "first that binds" rule in the spec anywhere? 19:41
19:41 wk__ joined
jnthn Yeah, S12 19:42
pmichaud didn't see it there but will look again
jnthn yes, just pulling latest
timotimo so, since junctions are already implemented in perl5, can someone tell me how the "leftmost and is autothreaded first" rule intermingles with named arguments? i'd like to write some tests for that and maybe implement it in rakudo as well
pmichaud okay, found it. 19:43
"for tiebreaker B the candidates are simply called in the order they were declared"
jnthn 1235 19:44
hoelzro this might shed some light on the subject: gist.github.com/hoelzro/5177985
jnthn oh, you meat me to it :)
pmichaud seems like the Any,Any candidate ought to be last, then, under the theory of proceeding from "most constrained" to "least constrained"
(in the source, that is)
jnthn I suspect that Any,Any will be nominally looser though.
pmichaud I know it doesn't matter in reality, but it might be easier to read/understand if the candidates were ordered that way in the source.
jnthn Yes, that is true 19:45
pmichaud I'll try that.
jnthn (Any,Any) is looser than both of (Real, Any) and (Any, Range).
But the latter two are tied.
So we fall back to bindability
With nameds counting as a constraint
Most common place nameds constrain is in trait mods. 19:46
19:46 Kurahaupo_ joined
pmichaud I suspect reordering will mean I can eliminate the Real,Whatever case, too. 19:46
jnthn arnsholt: Not SC_repossess_object. That's for something else 19:47
arnsholt: It's not owned at all, you just need to add it
arnsholt: Like is done with KnowHOW
arnsholt Thanks! 19:48
That's probably why I get a segfault =)
jnthn arnsholt: Also confused about the "don't have a name" comment. KnowHOWREPR.h shows a STRING *name to me :)
hoelzro would anyone mind taking a look at my gist? 19:51
arnsholt Indeed it does. I'm still a bit foggy on what types all the things actually are, so I was only looking at stuff common to all 6model objects =)
jnthn hoelzro: I'd expect that to give some error about EXPORT not expecting any args, but getting one 19:52
oh no
YOu put a |
hoelzro I thought that meant "take any args"? 19:53
jnthn yeah
I missed it on first read
hoelzro oh, ok
well, here's the error I'm getting:
Merging GLOBAL symbols failed: duplicate definition of symbol &EXPORT
arnsholt jnthn: Anyways, from looking at knowhow_bootstrapper.c all I have to do is set the ->sc member, no functions to call? 19:54
jnthn hoelzro: Try not making it "our"?
hoelzro jnthn: there we go!
jnthn hoelzro: ah, I thought I explained that and now I realize my explanation is useless without saying what UNIT is. 19:55
hoelzro: UNIT is the outermost lexical scope of something
It's what the module loader gets hold of when it loads a module, and it uses it to resolve things 19:56
Including finding your &EXPORT
arnsholt: No, there's pairings of lines:
VTABLE_set_pmc_keyed_int(interp, sc, 0, knowhow_pmc); SC_PMC(knowhow_pmc) = sc;
hoelzro alright
when I read that, I'd assumed that it mean the top level "chunk" of a file 19:57
arnsholt jnthn: Oh, right. I see, I see
jnthn arnsholt: The first of those puts the object into the SC. The 0 is the index.
19:57 SamuraiJack_ joined
arnsholt Yeah, so I'll probably want to VTABLE_push, rather than use a hardcoded index, right? 19:58
19:58 SamuraiJack left
hoelzro well, now that my EXPORT is being called...how do I inject routines into the calling context? 19:58
timotimo will i even be able to differentiate the positions of nameds between non-nameds from inside rakudo when i'm given a capture? :| 20:00
moritz timotimo: no 20:02
jnthn arnsholt: No becuase (1) I'm pretty sure the SCs don't work with that, and (b) we need to try and make the indexes match up between implementations in __6MODEL_CORE__ or we get problems. 20:03
hoelzro crap, now I need to figure out how to get errno via NativeCall =/ 20:04
timotimo moritz: in that case, should i write in the specs that all non-nameds and then all nameds will be considered separately, no matter if they are mixed? or should rakudo instead be able to differentiate those? 20:05
arnsholt hoelzro: I know the answer to that (unfortunately): You don't =(
hoelzro damn
I figured
arnsholt Unless there's a function that returns it, no way to get to global variables yet
hoelzro I'm guessing there's no Inline::C for Perl6?
FROGGS hoelzro: there is
hoelzro oh, good
moritz timotimo: the specs need to be adapted in some way
hoelzro I feel like I should just write a dynext for Parrot or something 20:06
but using NativeCall is probably "friendlier" to other impls
pmichaud I'd not be opposed to an errno opcode :)
ETOOMANYNEGATIONS "I think an errno opcode would be an okay shim for the time being, since it keeps coming up." 20:07
hoelzro pmichaud: should I add that to NQP? I was thinking of just writing a separate dynext that Rakudo could load
pmichaud hoelzro: I'd verify with jnthn first, but "add to NQP" is what I would think, yes. 20:08
hoelzro mkay
pmichaud nqp::posixerrno or something like that
arnsholt jnthn: Aha. So I need to be careful about the ordering of the boot_types 20:09
timotimo moritz: indeed. i opened a spec ticket a week ago or so and i'd like to advance that towards a solution. that's why i'm asking ;) 20:10
jnthn One thing with errno is we need to be careful about what may change it. 20:11
pmichaud jnthn: agreed, but I figure opening up the minefield will help us find out where the mines are. 20:12
or for a less gruesome analogy -- one way of finding the holes in the cheese is to let people start slicing it open :)
jnthn True :) 20:13
Was just wondering about making the native call stuff itself stash it away somewhere after each call
pmichaud right, we may end up wanting to do exactly that 20:14
jnthn And we expose an nqp::lasterrno() or so
Or something that makes clear the native call link
20:14 cognominal left
pmichaud or nqp::ncerrno 20:14
moritz which sounds decidedly non-threadsafe
jnthn +1
moritz: errno is thread safe itself afaikl; if our storage of it were to be thread local too it should be OK 20:15
hoelzro well, isn't errno already thread-specific?
pmichaud rn: say RangeIter
p6eval niecza v24-35-g5c06e28: OUTPUT«(RangeIter)␤»
..rakudo ce914f: OUTPUT«===SORRY!===␤Undeclared name:␤ RangeIter used at line 1␤␤»
jnthn RangeIter vanished from Rakudo as part of the immutable iterator changes, iirc? 20:17
moritz aye; Range became its own iterator
20:18 dvj joined 20:19 domidumont left
jnthn pmichaud: At some point in the not too distant future, it'd be interesting to discuss how to integrate the NQP JVM work into the nqp repo itself, and how we want to organize those kinds of things. 20:21
pmichaud: Not suggesting now, but I think we'll want it to happen in April.
pmichaud: So thought it'd be good to push it onto your mental stack :)
grondilu rn: .say for "AUAU".match: /^ A .* U .* /, :g 20:22
p6eval rakudo ce914f: OUTPUT«｢AUAU｣␤␤»
..niecza v24-35-g5c06e28: OUTPUT«Unhandled exception: Excess arguments to Cool.match, unused named g␤ at /home/p6eval/niecza/lib/CORE.setting line 0 (Cool.match @ 1) ␤ at /tmp/gM_RHAptut line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4299 (ANON @ 3) ␤ at /home/p6…
20:22 cognominal joined
grondilu rn: say .[0, 1] for "ACUGACU".match: /^ A (.*) U (.*) /, :g 20:24
p6eval rakudo ce914f: OUTPUT«｢CUGAC｣␤ ｢｣␤␤»
..niecza v24-35-g5c06e28: OUTPUT«Unhandled exception: Excess arguments to Cool.match, unused named g␤ at /home/p6eval/niecza/lib/CORE.setting line 0 (Cool.match @ 1) ␤ at /tmp/OUfloAX4JJ line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4299 (ANON @ 3) ␤ at /home/p6…
20:24 Rix joined
grondilu r: say .[0, 1].join: ":" for "ACUGACU".match: /^ A (.*) U (.*) /, :g 20:25
p6eval rakudo ce914f: OUTPUT«CUGAC:␤»
grondilu was expecting two matches
moritz the .* gobbles up everything 20:26
jnthn Try .*?
dalek kudo/nom: ff346ad | pmichaud++ | src/core/ (2 files):
Throw exception when Range endpoint is used to create a Range.
ast: 01d1162 | pmichaud++ | S03-operators/range (2 files):
Update spectests for Range endpoint exceptions.
grondilu r: say .[0, 1].join: ":" for "ACUGACU".match: /^ A (.*) U (.*) /, :overlap
p6eval rakudo ce914f: OUTPUT«CUGAC:␤»
grondilu r: say .[0, 1].join: ":" for "ACUGACU".match: /^ A (.*?) U (.*?) /, :overlap 20:27
p6eval rakudo ce914f: OUTPUT«C:␤»
moritz grondilu: and with ^ you can only ever get one match
(except with :exhaustive)
grondilu r: say .[0, 1].join: ":" for "ACUGACU".match: /^ A (.*?) U (.*?) /, :exhaustive 20:28
p6eval rakudo ce914f: OUTPUT«C:␤C:G␤C:GA␤C:GAC␤C:GACU␤CUGAC:␤»
dalek ecs: 2be5306 | pmichaud++ | S32-setting-library/Exception.pod:
(S32/Exceptions) Add X::Range::InvalidArg exception.
pmichaud jnthn: (nqp jvm) I figure you've given some thought to it already, so I'm fairly willing to follow your lead. By then I hope to be regularly building (and maybe hacking) on the jvm stuff, so I may be of more help then. 20:30
20:31 kurahaupo left
jnthn pmichaud: OK. Mostly just knowing things you consider desirable/unwanted in what I do there is what I'm after. 20:32
pmichaud: My current thought is to try and segregate backend-specific stuff into some src/backend/ 20:33
pmichaud: Thought maybe we want backend/jvm/...
So we can have src/, t/ (to segregate the tests also specific to each)
pmichaud segregate backend stuff +1 20:36
hoelzro if I were to add nqp_posix_errno, what would it do on Windows?
pmichaud hoelzro: whatever other languages/systems do for errno under windows 20:37
jnthn hoelzro: msdn.microsoft.com/en-us/library/5814770t.aspx
hoelzro got it
pmichaud also, prefer nqp_posixerrno for the Parrot opcode name, to correspond with nqp::posixerrno
(underscore consistency) 20:38
jnthn and perhaps more helpfully, msdn.microsoft.com/en-us/library/t3ayayh1.aspx
hoelzro ok, sounds good!
pmichaud I'm afk again... bbl or bbt 20:42
jnthn o/ 20:44
20:46 grondilu left 20:53 wk__ left
masak Rakudo release is on Thursday. we haven't assigned a release manager. 20:58
in fact, there aren't any release managers assigned for the upcoming releases.
let's make the regular deal -- I'll sign up for a release, if someone else steps forward and signs up for one. ;) 20:59
jnthn Sign up, sign up! You, yes you, could be the person who gets to relesae the first Rakudo release with JVM support!
masak I'll take it! :D
masak signs up for March 21:00
jnthn But you don't know which one it is!
Uh. :P
That's the gamble. You gotta pick a month. :)
hoelzro I assume NQP has a parallel release?
masak jnthn: I'm picking March!
jnthn Yes.
masak: :)
hoelzro is there a cut-off date for this month's release? 21:01
dalek kudo/nom: a323a31 | masak++ | docs/release_guide.pod:
[docs/release_guide.pod] grab March
jnthn masak: That's especially silly given you know that I have one week of Rakudo time before I have a week's mostly afk vacation :)
masak hoelzro: it's always on the Thursday after the third Tuesday on the month.
jnthn: what are you still doing standing here talkin'? get to work! :P
hoelzro masak: so as long as the change is in Git on Thursday, it goes out? 21:02
jnthn You want MOAR COMMITS from me?!
:P
masak jnthn: dude, I'm not asking you to do more than you're already doin'... :)
we're all so excited about the JVM, is all. 21:03
still no volunteer for April? sheesh, lurkers.
colomon \o/ \o/ \o/
dalek kudo/nom: c22a391 | jnthn++ | src/Perl6/ModuleLoader.pm:
More pir:: => nqp:: in ModuleLoader.
 21:04
kudo/nom: d00a575 | jnthn++ | src/Perl6/World.pm:
A couple of pir:: => nqp:: in Perl6::World.
kudo/nom: e7f9bdd | jnthn++ | src/Perl6/Compiler.nqp:
Last pir:: in Perl6::Compiler.
kudo/nom: d98dd54 | jnthn++ | src/Perl6/Metamodel/ (7 files):
Some pir:: => nqp:: in the MOP.
arnsholt I'm not gonna sign up for April =)
It'd be cool, but with teaching duties and article deadline in late April, I don't think it'd be prudent =)
masak .oO( un-vol-teer )
arnsholt Neat one!
masak I'll etym your ology! 21:05
sorear good *
arnsholt =D
'lo sorear
nwc10 how much pir:: is left in Rakudo? And how much C?
masak sorear! \o/ 21:06
arnsholt Not a lot of PIR I think
IIRC most of the C code was multi-dispatch as well =)
arnsholt makes a pot of tea 21:07
jnthn nwc10: Sadly, numerical metrics aren't too infornative here.
nwc10: Some things are much easier to make portable than others.
One big problem that arnsholt is about to run straight into, that also needs solving for the JVM stuff, is that we've used "that's a Parrot type, not a 6model one" to differentiate the NQP/Rakudo boundary in a bunch of places. 21:08
21:08 wk_ joined
arnsholt Oh. That's no fun =/ 21:08
jnthn And over in JVM land, every object we touch is a 6model one. And NQP on Parrot is getting more that way once we do the NQPArray thing.
So that needs solving properly before the serious porting efforts go on. 21:09
arnsholt Yeah. I'm hoping once I get this VMArray stuff working, moving over the rest of the NQP* stuff wil mostly be a question of implementing the REPRs
jnthn Container stuff needs a look, but that's been the case for a long while, not just because of the JVM stuff. Of note, the whole "is rw" with natives interaction. 21:10
gather/take needs a LOT of distilled alco^W^W^W^Wbit of careful effort to do the rest of the continuations through exceptions as a stack iterator mechanism. 21:11
And there's still small matter of the binder and how the heck we can try to organize things there to be able to take advantage of invokedynamic.
The last one of those aside, neither of these equate to huge numbers of pir:: or lines of C code, but they're the conceptually tricky areas. 21:12
21:14 MayDaniel left
nwc10 yes. they sound it. thanks for that explanation 21:15
jnthn np :)
None of them are things I look at and think "omg, how on earth can it be done". :) But none of them are "yeah, I'll nail it in an afternoon" either :)
nwc10 yes, they all sound do-able. But big 21:16
21:18 Kurahaupo_ left, kurahaupo joined
arnsholt jnthn: Does that mean the vmarray stuff should live in its branch until you figure that stuff out properly? 21:20
Oooh, this time the compilation gets to stage2 before crapping out \o/ 21:21
jnthn arnsholt: I suspect at least the NQPArray aspects of it will have to 21:25
arnsholt Right, right
Any pointers on how to track down a Parrot-level ("Null PMC access in assign_pmc()") error, BTW? 21:26
jnthn Not besides looking at the stack trace. 21:27
That's a slightly unusual thing to be doing, though.
(Not many things do the assign PMC thing0
arnsholt Thinking for a bit I think it's in my new nqp::list, since it does: $ops.push_pirop('assign', $list_reg, $arr); 21:28
Of course, that may be entirely the wrong thing
jnthn Should be set
not assign 21:29
arnsholt Right. I wasn't quite sure which one was right from looking at the opcode descriptions
jnthn well, I'm guessing without context, but it seems a decent guess :)
arnsholt Now that you mention set I remember being unsure about the two, so it seems a plausible option 21:30
Any objections to removing qlist entirely, BTW? 21:31
dalek pan style="color: #395be5">perl6-examples: 9ae6e4c | (L. Grondin)++ | rosalind/dna-grondilu.pl:
[rosalind] DNA
arnsholt It's not used at all in NQP AFAICT, and only two occurences (in one file) in Rakudo 21:32
21:36 grondilu joined
jnthn I suspect its usage in those places is important for efficiency. 21:36
I'd leave it for now and we remove it once we get NQPArray in place; VMArray uses the QRPA storage strategy so it should be fine for all the usages. 21:37
arnsholt Yeah, that's what I thought
I figured I'd remove it once I get the nqp::list stuff working (since then nqp::list will operate like a QRPA) 21:38
But first, figure out why I get Null PMC in push 21:39
21:40 bruges left
arnsholt tells the compiler to insert some print statements 21:40
21:41 bruges joined 21:43 quester left
arnsholt jnthn: It seems my nqp::list returns a null PMC 21:43
jnthn Do you set a .result on the PIRT::Ops at some point? 21:45
arnsholt Oh, that's probably what I'm doing wrong 21:46
I do "$ops.push_pirop('set', $list_reg, $list);", but list is just "$qastcomp.as_post(QAST::Op.new(:op('list')));" 21:47
I guess it should be $list.result?
jnthn The result is the register that the overall set of operations should evaluate to 21:49
arnsholt Hang on, lemme paste you the code 21:51
21:52 domidumont joined 21:53 kaare__ left
arnsholt jnthn: gist.github.com/arnsholt/5178485 21:53
nqp::hlllist just returns the BOOTArray type for the time being
The Null PMC is in the push in list_b 21:54
jnthn arnsholt: You never do $ops.push($arr) 21:55
At least, not before you try to use the result
arnsholt Oh, derp
jnthn :)
arnsholt Won't work unless I actually execute it, will it? =D 21:56
jnthn Suspect not ;-)
arnsholt "elements() not implemented in class 'SixModelObject'" \o/ 22:02
Right. About time to make some sort of commit, I think 22:03
jnthn: Do you have any specific wants for how sethllconfig should be implemented on NQP/Parrot, BTW? 22:05
jnthn arnsholt: I think there needs to be a similar mechanism to on the JVM. The subtlety in it all is really timing, I discovered... 22:06
arnsholt: You only want to create the underlying state for a HLL once. Even if that first time is a request for it. 22:07
arnsholt: So the sethllconfig may be updating something already existing.
arnsholt: That's 'cus other things may take direct references to it.
arnsholt Right 22:08
pmichaud /msg p6eval r: say 1 .. ^10
jnthn arnsholt: The stuff I've got sketched in my notebook (but I need to sanity review it) was that we'll let STables carry a "which HLL owns this type" and that we'll use that
pmichaud: fail!
pmichaud fail :-)
r: say 1 .. ^10 22:09
p6eval rakudo d98dd5: OUTPUT«Range objects are not valid endpoints for Ranges␤ in method new at src/gen/CORE.setting:5297␤ in method new at src/gen/CORE.setting:5289␤ in sub infix:<..> at src/gen/CORE.setting:5480␤ in block at /tmp/CrPoRAfcBQ:1␤␤»
pmichaud \o/
jnthn pmichaud++
arnsholt: uh, that we'll use that for answering the questions we answer with "it's a non-6model thing" today.
arnsholt Right, right. That sounds like it makes sense 22:10
masak pmichaud++
jnthn This should also help me deal with the various other nasties when NQP objects leak into Rakudo land.
r: say Sub.^methods
p6eval rakudo d98dd5: OUTPUT«No such method 'gist' for invocant of type 'Sub'␤ in method gist at src/gen/CORE.setting:5165␤ in method gist at src/gen/CORE.setting:902␤ in sub say at src/gen/CORE.setting:7634␤ in block at /tmp/oov_9VMT9t:1␤␤»
jnthn That one, for example.
arnsholt Yeah, explicitly keeping track of that sounds like a good idea 22:11
jnthn pmichaud: One interesting consequence of developments in the last few months is that we can make Parcel know to delegate positional operations to an underlying storage 22:13
pmichaud: So we'll be able to nqp::atpos(...) directly on the Parcel, which'll clean up some code. 22:14
22:14 SamuraiJack_ left 22:23 domidumont left 22:27 Exodist joined, census left 22:29 census joined
arnsholt jnthn: Is there a principle which decides which vtables in sixmodelobject.pmc need to decontainerize their SELF first? 22:30
22:30 rindolf left
jnthn arnsholt: All but those doing attribute lookup/binding. 22:31
(get_attr/set_attr)
arnsholt Excellent. Thanks! 22:32
masak just to repeat: 'decontainerize' means "strip away the container and look at the underlying value"?
jnthn Yes
Spoiler: I'm going to incorporate the assignment bit of things into the protocol container spec stuff too. 22:33
s/protocol //
arnsholt I must admit decontainerization is still somewhat magical to me, still. I just know it's important to do =) 22:34
jnthn r: my $a = 42; my $b := 42;
p6eval rakudo d98dd5: ( no output )
jnthn In the above, $a in the lexpad contains a Scalar, which we put at 42 in
By contrast, $b points straight to 42 22:35
When we do $a.sin, we mean the thing inside of the container.
Decontainerization makes sure we remove the container. 22:36
22:39 yoleaux left, yoleaux joined
arnsholt That's odd. It says that "get_bool() not implemented in class 'SixModelObject'", but it's clearly there... 22:39
Unless that's the error you get if the 6model thingy does say how to be boolified 22:41
masak I've always felt there's something odd about the word "Scalar" there. seems like a category error of sorts.
now I think I know what it is. it should be called "Item" or "Container". 22:42
the scalar (a string or a number or a reference to something) is what you store *inside* the container.
arnsholt jnthn: Next time we meet, remind me to buy you a beer =) 22:43
dalek rl6-roast-data: 1d939a0 | coke++ | / (4 files):
today (automated commit)
 22:47
masak .oO( The Perl 6 community is a machine that turns volunteers into beer for jnthn, who turns it into commits ) 22:49
arnsholt Indeed
jnthn
.oO( I thought beer was made out of malt and hops and stuff... )
arnsholt jnthn++ # Commits and stuff 22:50
Soylent beer is people!
dalek rl6-most-wanted: 6bf2646 | (Pawel Pabian)++ | most-wanted/bindings.md:
Added GeoIP.

Supports paid and free Geo City databases from MaxMind. Tested under OS X and Ubuntu Linux.		 22:51
arnsholt Only tangentially related to anything, I love how git makes it easy to speculatively hack away on something, and then pack it into sensible commits later
masak git packages the insight that learning happens during development into commands that help you factor that learning into (private) history. 22:55
'night, #perl6
jnthn Git is very good at letting you bring order to naturally messy/explorative development :)
arnsholt 'night masak 22:56
See you tomorrow
jnthn o/ masak
masak .oO( not if I accidentally mail you private correspondence first! )
zzz & 22:57
23:06 PacoAir left
jnthn [Coke]: I'm curious about the two new failures 23:06
23:10 yoleaux left, yoleaux joined, yoleaux left 23:11 yoleaux joined 23:12 spider-mario left
arnsholt jnthn: "get_iter() not implemented in class 'SixModelObject'", I guess this means I get to implement VMIter as well? =) 23:18
jnthn Oh my... :)
Well, things are certainly all set up so doing things that way will work, thanks to the fact NQP-JVM does that. 23:19
arnsholt Except VMIter might not speak the same iteration protocol as Parrot expects? 23:20
jnthn It's close enough
Note that NQPCORE in JVM defines a little glue to make things work out, especially for the Hash iteration 23:21
And the MOP is factored in terms of ops that don't expect that sugar
23:21 cognominal left
arnsholt The stuff in NQPCORE being ArrayIter and HashIter? 23:23
jnthn yes
Essentailly, the stuff over in NQPJVM defines a 6model factoring of pretty much everything NQP uses 23:24
arnsholt Which makes sense, now that we're targeting multiple backends 23:26
23:28 diakopter joined
jnthn Yeah. It's very much closer to "how to do it". 23:28
I'm happy to have NQP on Parrot converge that way.
23:29 cognominal joined 23:30 diakopter left, diakopter joined
arnsholt So, this is fun. parrot/src/pmc/arrayiterator.c claims to be generated from parrot/src/pmc/arrayiterator.pmc. Of course, that file doesn't exist 23:32
jnthn huh, it does here... 23:33
arnsholt Maybe I did something weird 23:35
23:36 jaldhar left
arnsholt Apparently git has never heard of the file... 23:36
23:36 _jaldhar joined
jnthn arnsholt: Knowte that the parrot dir is a different repo 23:37
arnsholt Maybe I've done something odd to my parrot tree
jnthn uh, note
...weird typo
arnsholt Yeah, this is while inside the Parrot repo
jnthn oh. Then...weird
arnsholt I think I'll try with a fresh checkout 23:38
Whoa, that was quick. rm -rf parrot/ install/ usually takes a while. Yay SSD! =)
jnthn: NQP/JVM has an op called iter, while NQP/Parrot has iterator. Intentional or typo? 23:41
jnthn QAST::OperationsJAST.map_classlib_core_op('iterator', $TYPE_OPS, 'iter', [$RT_OBJ], $RT_OBJ, :tc); 23:42
It ended up getting called iter in Ops.java somehow
But it's exposed as nqp::iterator
arnsholt Oh, right. Never mind me, then =)
jnthn Feel free to consistentize it. :) 23:43
23:43 cognominal left
arnsholt It's consistent with Parrot though =) 23:43
The native pirop is called iter, after all
jnthn ah, and if I added it when porting 'for'... 23:44
23:44 dpk joined
arnsholt There. Now I have arrayiterator.pmc 23:50
23:51 _jaldhar left, census_ joined 23:52 census_ left 23:53 census left 23:54 census joined, _jaldhar 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!