»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' or rakudo: / pugs: / std: , or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by diakopter on 25 January 2010.
TimToady it again has the feel of something being introspected that should Just Work... 00:00
jnthn Well, it usually does.
Well 00:01
TimToady there's some kind of multi-ish thing that's getting dispatched to that is outside the role itself, maybe
jnthn If I add role R { } to my collection...
then I can do class Foo does R { } and it really means class Foo does R[] { }
But I've not got a good way to write trait_mod:does to know that it should take R (the general thing) and actually resolve it to a particular role compose R[]. 00:03
TimToady as in, call the hidden R-doer routine with ()
something curry-ish, like .assuming() 00:04
jnthn I'm not sure that helps, given that if somebody wrote class Foo does R[Int] { } then by the time we get to the trait-mod, the thingy that makes it specialized has alreaedy happened.
TimToady only that doesn't limit to ()
I'm probably still missing something about what you're trying to do that is the immediate cause 00:05
jnthn Well, that dispatch thingy kinda *is* the thing I'm talking about I guess, when I refer to the group of all roles under the name R.
OK, let me step back and explain the current model I have. 00:06
When I wrote: role R { }; role R[::T] { }; role R[::T1, ::T2] { } then one thing gets installed in the namespace/lexpad under R. It knows about all of the variants, and internally contains a Multi that it just calls to pick the variant. 00:07
00:07 k23z__ joined
jnthn Each of those multis serves as a kinda "role factory", and uses RoleHOW to build up a role, and through closure semantics it's methods end up seeing the current parameters given to the role. 00:08
Suppose I write:
class Foo does R { }; class Foo does R[Int] { } 00:09
I have a trait_mod:does, but in one case it has R - the thing that knows about all of the roles. What we *want* is to compose R[] - the first parameterless variant. In the second case, we already did that, and we have a role ready to compose. 00:10
My problem is what trait_mod:does should look like here.
We could say that doing [] on any role (e.g. not the factory of them) is just identity.
00:10 colomon joined
TimToady zennishly 00:10
jnthn And trait_mod:does just always does that to what it's given. 00:11
TimToady well, [] is more like () then a zen slice, so an explicit [] would explicitly mean no args
jnthn TimToady: Yeah. 00:12
We could pull it off that way.
The trickier bit is that we probably want R ~~ Role and R[Int] ~~ Role
TimToady I suppose we could say that any role that has args must be used with [], and R *always* means the un-called multi
and blow up if you say 'class Foo is R {...}' 00:13
jnthn Yeah, but do we really want to make people write class Foo does R { }
TimToady with some appropriate message
they'd have to say class Foo does R[] if R has parameters
jnthn I mean, parametric roles are useful, I'm just not convinced they're the Common Case.
"has parameters"?
There may be a role R { ... } and a role R[::T] { ... } 00:14
TimToady perhaps an error, and you have to say role R[] for the first one
could even require 'multi' keyword
jnthn It'd resolve the ambiguity perhaps, but I'm not sure if I like forcing people to write [] after all their role names when they do them. 00:15
TimToady only the ones that want parameters
jnthn Ah, so
TimToady does Positional is still just fine
maybe
bad example perhaps
jnthn Well, Positional is role Positional[::T = Mu] { ... } I guess. 00:16
TimToady anyway, does can probably do something special in the context of requiring an instantiated role, so it can assume [], whereas bare rvalue R is always the multi name, maybe 00:21
00:22 kensanata left
jnthn Something like that probably works. 00:23
Should both R ~~ Role and R[Int] ~~ Role be true?
00:24 pmurias left
TimToady probably, but one of them could also respond to a more specific type 00:25
AbstractRole, if it's the one that needs [] as rvalue 00:26
or some such
00:26 eternaleye left
jnthn Yay 00:26
Knowing that this is reasonable helps a lot.
Thanks.
TimToady or maybe ConcreteRole for the other :)
jnthn :-) 00:27
TimToady ReadyToRole
jnthn :D
TimToady hides in shame
00:27 stephenlb left
jnthn It's OK, telling people to "Role up, role up" to one of my talks on roles was just as shameful. :-) 00:27
TimToady HardRoles :) 00:28
CastRoles 00:29
to go back to the original metaphor 00:30
well, it's the class that is the actor, really, so that doesn't work
00:31 stephenlb joined
TimToady it's more like Bill does Clown vs Bill does Bozo 00:31
Bill does Clown[Bozo] I should say 00:32
jnthn :-)
00:32 bbkr left, bbkr joined 00:33 go|dfish left
TimToady Comic: that's one Bill too many... 00:33
00:34 andy2 joined
TimToady Sidekick: no, duck typing is one Bill too many... 00:34
Comic slugs Sidekick, who responds "Mu"... 00:35
TimToady 's arm and fist both hurt now. 00:37
00:39 eternaleye joined 00:45 go|dfish joined 00:46 eternaleye left
TimToady apologizes for all the wise-quacks, and ducks. 00:49
colomon jnthn++: class Whatever { ... } works great in operators.pm now. Danke
00:49 eternaleye joined
jnthn Bitte. 00:50
00:52 PZt joined 00:53 cognominal left 00:54 cognominal joined 00:56 orafu joined
jnthn colomon: BTW, does the multi fix unblock getting various trig bits back? 01:03
colomon It should be a good start, anyway. I'll try to take a proper look at it soon. 01:04
I was thinking I'd try to get more Range tests working first, then look at getting Rats back in shape, then trig.
jnthn Sounds like a good plan. 01:05
dalek kudo/ng: cf189ea | (Solomon Foster)++ | src/core/operators.pm:
Start at a infix:<...>(@lhs, Whatever) implementation, which utterly fails to be called at the moment.
 01:06
01:10 cdarroch left 01:11 cl58 joined
cl58 needs moritz_ 01:11
moritz_: ping (come back from away!)
01:15 eternaleye left 01:18 drbean joined 01:23 eternaleye joined 01:24 johnz joined 01:25 ihrd joined, mssm left
jnthn OK, sleep time for me... o/ 01:33
colomon \o 01:41
01:44 synth joined 01:45 andy2 left
colomon ng: my @a = 1, 4; (@a ... *).GrabAndSay(10); 01:48
p6eval ng cf189e: OUTPUT«sh: ./perl6: No such file or directory␤»
colomon ng: my @a = 1, 4; (@a ... *).GrabAndSay(10); 01:50
p6eval ng cf189e: OUTPUT«sh: ./perl6: No such file or directory␤»
dalek kudo/ng: 0af588e | (Solomon Foster)++ | src/core/operators.pm:
Add two more cases for the series operator.
 01:52
02:03 Lorn left, ihrd left 02:04 lichtkind left
pugs_svn r29669 | lwall++ | [STD] delete old p5=> that masak++ noticed 02:07
colomon ng: my @a = 1, 4; (@a ... *).GrabAndSay(10);
p6eval ng cf189e: OUTPUT«1␤4␤7␤10␤13␤16␤19␤22␤25␤28␤»
colomon ng: my @a = 1, 4, 16; (@a ... *).GrabAndSay(10);
p6eval ng cf189e: OUTPUT«Method 'GrabAndSay' not found for invocant of class 'Integer'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon ng: my @a = 1, 2, 4; (@a ... *).GrabAndSay(10); 02:08
p6eval ng cf189e: OUTPUT«Method 'GrabAndSay' not found for invocant of class 'Integer'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady colomon: note that .batch(10) is already specced to do the grabbing, though not the saying
colomon ah, it hasn't caught up with my latest patch yet.
TimToady: that's Iterator.batch? 02:09
TimToady yes
02:09 lest_away is now known as lestrrat
colomon TimToady: hmmm.... 02:09
02:10 meppl left
colomon ng: my @a = 1, 2, 4; (@a ... *).getbatch(10).perl.say 02:11
p6eval ng cf189e: OUTPUT«Method 'getbatch' not found for invocant of class 'Integer'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon ng: my @a = 1, 4; (@a ... *).getbatch(10).perl.say
p6eval ng cf189e: OUTPUT«(28, 4, 7, 10, 13, 16, 19, 22, 25, 28)␤»
colomon umm..... 02:12
ng: my @a = 1, 4; (@a ... *).GrabAndSay(10);
p6eval ng cf189e: OUTPUT«1␤4␤7␤10␤13␤16␤19␤22␤25␤28␤»
02:12 lue joined
TimToady I think I see how that went wrong :) 02:13
colomon TimToady: is "take $i.WHICH" an official way to return the value rather than the variable itself, or just a notion?
TimToady a value type always returns its value for WHICH 02:14
it wouldn't work for objects
02:14 Austin joined
colomon Thank you. 02:15
02:15 cl58 left, hercynium left
colomon was .getbatch an earlier name for .batch, or is there a subtle difference? 02:15
02:15 cl58 joined, Austin left
TimToady yes, decided batch was verby enough for the other dimension from '' vs 'obj' 02:16
where '' is short for 'flat' :)
so 'get' is really getflat, and batch is batchflat 02:17
and 'obj' is short for 'lumpy' :) 02:18
still wonder whether we should have gone for smooth and crunch instead... 02:19
*crunchy
dinner & 02:20
dalek kudo/ng: 63cafa9 | (Solomon Foster)++ | src/core/operators.pm:
Fix to infix:<...> so that it returns the first value rather than a reference to the loop variable.
 02:22
lue Just out of curiosity, why is the bot named 'dalek'? 02:23
eternaleye lue: Someone likes Dr. Who? 02:24
lue what I was thinking. But it makes me scared on ng a little :) 02:25
'of', not 'on'
If I ever push a commit to rakudo/ng, I'll make the comment begin with EXTERMINATE! 02:29
colomon rakudo: say ~("Y".."AB") 02:55
p6eval rakudo 1d4928: OUTPUT«Y Z AA AB␤»
colomon ng: say ~("Y".."AB") 02:56
p6eval ng 9fa0fd: OUTPUT«␤»
02:58 hercynium joined 02:59 JimmyZ joined 03:06 cognominal left 03:09 ShaneC left 03:14 stephenlb left 03:16 zorgnax left 03:27 cotto_working left 03:28 cotto_working joined
colomon is wondering what it means that his last push has not been announced here, nor does it show up in the commits list on github. 03:29
ng: say 1
p6eval ng 9fa0fd: OUTPUT«1␤»
colomon looks like p6eval has it, however, so I guess it must have left my laptop.
03:29 LylePerl left
colomon ng: my @a = 1, 2, 4; (@a ... *).batch(10).perl.say 03:29
p6eval ng 9fa0fd: OUTPUT«(1, 2/1, 4/1, 8/1, 16/1, 32/1, 64/1, 128/1, 256/1, 512/1)␤»
colomon ng: my @a = 4, 2, 1; (@a ... *).batch(10).perl.say 03:30
p6eval ng 9fa0fd: OUTPUT«(4, 2/1, 1/1, 1/2, 1/4, 1/8, 1/16, 1/32, 1/64, 1/128)␤»
colomon ng: say 1, 2, 3 Z 4, 5, 6 03:43
p6eval ng 9fa0fd: OUTPUT«Could not find non-existent sub &infix:<Z>␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon ng: (0..3).map({$_ * 2}).join('|') 03:47
p6eval ng 9fa0fd: ( no output )
colomon ng: (0..3).map({$_ * 2}).join('|').say
p6eval ng 9fa0fd: OUTPUT«0|2|4|6␤»
JimmyZ ng: ((4, 2, 1) ... *).batch(10).perl.say 03:52
p6eval ng 9fa0fd: OUTPUT«Method 'succ' not found for invocant of class 'Seq'␤current instr.: '&prefix:<++>' pc 313961 (src/gen/core.pir:40849)␤»
colomon JimmyZ: something's wrong with the grammar (I think) which blocks the straightforward forms calling of infix:<...> 03:53
JimmyZ aye 03:55
colomon ng: (1..3).first({ $_ % 2 == 0}).say 04:01
p6eval ng 9fa0fd: OUTPUT«1␤»
colomon ng: (1..3).first({ $^a % 2 == 0}).say
p6eval ng 9fa0fd: OUTPUT«2␤» 04:02
colomon ng: my @one = 1; say ~(@one .. 3); 04:04
p6eval ng 9fa0fd: OUTPUT«Method 'succ' not found for invocant of class 'Array'␤current instr.: '!dispatch_.=' pc 440 (src/glue/dispatch.pir:118)␤»
pugs_svn r29670 | colomon++ | [t/spec] Fudge for ng. 04:05
04:07 cl58 left
dalek kudo/ng: 9fa0fd1 | (Solomon Foster)++ | src/ (2 files):
Rename Iterator.getbatch to Iterator.batch.
 04:10
kudo/ng: 1feef8c | (Solomon Foster)++ | t/spectest.data:
Turn on S03-operators/range.t.
TimToady ng: say (0..*)[^10].perl 04:12
p6eval ng 9fa0fd: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class 'Range'␤current instr.: '!postcircumfix:<[ ]>' pc 314127 (src/gen/core.pir:40943)␤»
TimToady there's another version of grab-the-first-10
JimmyZ ng: (<4, 2, 1> ... *).batch(10).perl.say 04:15
p6eval ng 9fa0fd: OUTPUT«Method 'succ' not found for invocant of class 'Seq'␤current instr.: '&prefix:<++>' pc 313961 (src/gen/core.pir:40849)␤» 04:16
TimToady don't need the commas
04:17 lestrrat is now known as lest_away, jaldhar joined
JimmyZ ng: (<4 2 1> ... *).batch(10).perl.say 04:20
p6eval ng 9fa0fd: OUTPUT«sh: ./perl6: No such file or directory␤»
JimmyZ ng: (<4 2 1> ... *).batch(10).perl.say 04:21
p6eval ng 9fa0fd: OUTPUT«sh: ./perl6: No such file or directory␤»
TimToady recompiling, but probably won't anyway
won't work
JimmyZ maybe
TimToady unless someone has made <4 2 1> autoconvert to integer already
JimmyZ ng: ((4, 2, 1) ... *).batch(10).perl.say
p6eval ng 1feef8: OUTPUT«Method 'succ' not found for invocant of class 'Seq'␤current instr.: '&prefix:<++>' pc 313961 (src/gen/core.pir:40999)␤»
JimmyZ ng: (<4 2 1> ... *).batch(10).perl.say
p6eval ng 1feef8: OUTPUT«Method 'succ' not found for invocant of class 'Seq'␤current instr.: '&prefix:<++>' pc 313961 (src/gen/core.pir:40999)␤»
TimToady looks like ... isn't actually behaving like a list infix 04:22
JimmyZ ng: my @a = 1, 2, 4; (@a ... *).batch(10).perl.say 04:23
p6eval ng 1feef8: OUTPUT«(1, 2/1, 4/1, 8/1, 16/1, 32/1, 64/1, 128/1, 256/1, 512/1)␤»
JimmyZ yes
ng: my @a = 4, 2, 1; (@a ... *).batch(10).perl.say 04:24
p6eval ng 1feef8: OUTPUT«(4, 2/1, 1/1, 1/2, 1/4, 1/8, 1/16, 1/32, 1/64, 1/128)␤»
TimToady maybe Seq isn't flattening correctly
04:24 k23z__ left
TimToady but also, <4 2 1> is probably still making strings 04:24
04:25 justatheory left
TimToady (1, 2, 4, ... *).[^10].perl.say 04:27
ng: (1, 2, 4, ... *).[^10].perl.say
p6eval ng 1feef8: OUTPUT«No exception handler and no message␤current instr.: '&fail' pc 15242 (src/builtins/Junction.pir:207)␤»
TimToady ng: (1, 2, 4 ... *).[^10].perl.say
p6eval ng 1feef8: OUTPUT«Method 'succ' not found for invocant of class 'Seq'␤current instr.: '&prefix:<++>' pc 313961 (src/gen/core.pir:40999)␤»
TimToady ng: ((1, 2, 4).flat ... *).[^10].perl.say 04:28
p6eval ng 1feef8: OUTPUT«Method 'flat' not found for invocant of class 'Parcel'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady ng: (1, 2, 4 ... *).Array.[^10].perl.say 04:31
p6eval ng 1feef8: OUTPUT«Method 'Array' not found for invocant of class 'GatherIterator'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
sjohnson rakudo: say 'hi'; 04:32
p6eval rakudo 1d4928: OUTPUT«hi␤»
sjohnson rakudo: say ord('h'); 04:33
p6eval rakudo 1d4928: OUTPUT«104␤»
eternaleye I think it's that there is no infix:<...>( Seq $lhs, Whatever )
ng: say Seq.new( 1, 2 ).flat
04:34 gfx joined
p6eval ng 1feef8: OUTPUT«Method 'flat' not found for invocant of class 'Seq'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤» 04:34
eternaleye ng: say Seq.new( 1, 2 ).Array
p6eval ng 1feef8: OUTPUT«Method 'Array' not found for invocant of class 'Seq'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
eternaleye ng: say Seq.new( 1, 2 )
p6eval ng 1feef8: OUTPUT«Null PMC access in get_integer()␤current instr.: 'perl6;Seq;elems' pc 265145 (src/gen/core.pir:23332)␤»
eternaleye o.0
TimToady shouldn't need that exact signature
eternaleye I wonder if that null PMC error is known
TimToady in fact, list infixes should generally have a sig more like ([*@lhs], [*@rhs]) or some such that will put flat context on both sides 04:35
eternaleye ng: say Seq.new( 1, 2 ).^methods( :local )>>.name.join(', ') 04:36
p6eval ng 1feef8: OUTPUT«Null PMC access in get_integer()␤current instr.: 'perl6;Seq;elems' pc 265145 (src/gen/core.pir:23332)␤»
eternaleye ng: say Seq.^methods( :local )>>.name.join(', ')
p6eval ng 1feef8: OUTPUT«Null PMC access in get_integer()␤current instr.: 'perl6;Seq;elems' pc 265145 (src/gen/core.pir:23332)␤»
TimToady we don't want to have to write signatures for every single type that might be asked for a list 04:37
eternaleye Gah, I'm channeling the Null PMC Bunny
ng: say ~[Seq.new( 1, 2 )] 04:38
p6eval ng 1feef8: OUTPUT«Null PMC access in get_integer()␤current instr.: 'perl6;Seq;elems' pc 265145 (src/gen/core.pir:23332)␤»
TimToady so probably you just want infix:<...> (Any, Any) with appropriate coercions internally 04:39
or maybe list infixes automatically just get two parcel objects 04:40
eternaleye The latter sounds cleanest 04:41
TimToady more like a 2D slice, really
and extends to list-associative listops
eternaleye But can signatures for the contents of those parcels be specified? 04:42
TimToady probably
**$left, **$right to begin with 04:43
eternaleye multi sub infix:<...>( :( Numeric $a, Numeric $b ), :( Whatever ) ) might be an interesting syntax
Using the signature decl syntax 04:44
TimToady maybe. at some point it's easier to just pull in the lists and look at the values
eternaleye BTW, if ** is a singleton Whatever, what's **? 04:45
erm, s:1st/\*\*/*/
TimToady HyperWhatever, but that's not what it means in a signature 04:46
it takes over from the old @@
it's "slicey" instead of "slurpy"
eternaleye Yep. I was asking abut it in term position. My brain jumps around a lot. 04:47
TimToady and 2D instead of 1D
as rvalue generally means you can replace multiple dimensions at that spot
sorta [;] * xx *
except we don't have [;] anymore 04:48
but maybe removing [;] was ill-considered
now it's **(@dimensions) 04:53
and ** is short for **(* xx *)
eternaleye Sweet 04:54
TimToady arguably it should be something like || to go with |
||(* xx *) 04:55
though prefix || is gonna look funny to a C programmer
eternaleye Or someone who has written ebuilds on Gentoo 04:56
There, ||( ) in the dependency spec means 'any of these is okay, switchable at will' 04:57
TimToady I'm thinking slice interpolation is likely to change to prefix:<||>, since it's really the 2D form of prefix:<|> 04:59
also, as a prefix wouldn't require ()
eternaleye Makes sense
TimToady ||@foo
and the reduplication means the same thing for */**
eternaleye Plus, it would preserve the |prefix feeding *signature symmetry, with ||prefix feeding **signature 05:00
TimToady well, I'll think about it while I pick up kid from evening school. 05:01
bbl &
05:04 hanekomu left 05:07 hanekomu joined 05:21 cjk101010 joined
pugs_svn r29671 | lwall++ | [S06,S09] change **() special form to prefix:<||> by analogy to prefix:<|> 05:36
r29672 | lwall++ | [STD] allow prefix:<||> 05:41
05:51 agentzh joined
pugs_svn r29673 | lwall++ | [S03] add description of prefix:<||> semantics 05:52
05:53 lest_away is now known as lestrrat 05:59 quietfanatic joined 06:01 lue left 06:03 synth left 06:04 [particle] joined 06:26 agentzh left, agentzh joined 06:47 patspam left 06:49 fridim joined 06:54 szabgab joined 07:07 cognominal joined, kaare joined, kaare is now known as Guest55887 07:09 am0c left 07:15 uniejo joined, JimmyZ left 07:17 nabax joined 07:19 Su-Shee joined
Su-Shee good morning 07:20
07:23 diakopter sets mode: +o Su-Shee
diakopter in case le spammers strike 07:23
griefers 07:24
zzzzzzzzzzzzzzzzzzzzzzzzzzzzz&&&&
07:28 eternaleye left
sjohnson diakopter: i think the new ircd has eliminated that problem 07:28
i think it has better spam detection technology 07:29
spinclad TimToady: i don't think i would call unary '||' a _flattening_ operator, as i think of flattening as what slurpy context does, as opposed to slice context 07:39
07:39 eternaleye joined
spinclad (re last edit to S03) 07:40
szbalint freenode staff aren't very good in handling security/spam :) 07:42
spinclad they can't be everywhere at once
07:47 iblechbot joined 07:53 drbean left 08:02 rv2733 joined 08:16 kst left 08:24 lestrrat is now known as lest_away
eternaleye TimToady: You said [;] had been nuked, but S32/Containers has an invocation of it under cat() 08:25
08:27 lest_away is now known as lestrrat, clausi joined
frettled eternaleye: some roaches survive nuking ;) 08:29
eternaleye Just wanted to point it out.
So the Giant Foot of Squishing could get it 08:30
frettled I hope the Giant Foot of Squishing has a Giant Boot of Protection, that looks mighty sharp. ;)
mathw The Giant Foot of Squishing? I like that
I would imagine it probably has a good few levels of Heavy Armour 08:31
08:34 cosimo left 08:42 cosimo joined 08:48 agentzh left 08:50 rgrau joined
moritz_ good morning 08:52
phenny: tell cl58 it seems we don't share much common uptime these days - feel free to /msg me, or send an email to [email@hidden.address] 08:53
phenny moritz_: I'll pass that on when cl58 is around.
moritz_ phenny++
08:54 mj41 left, Speedy1 joined
Speedy1 www.search2.net 08:54
mathw morning moritz_ 08:55
08:55 rgrau left 08:57 mj41 joined 08:58 Speedy1 left 09:06 nabax left 09:10 umut joined 09:11 cosimo left 09:12 fridim left 09:14 umut left, cosimo joined 09:17 dakkar joined 09:21 fridim joined 09:25 lestrrat is now known as lest_away 09:29 IllvilJa left 09:40 payload left 09:46 IllvilJa joined
moritz_ hugme: list projects 10:02
hugme moritz_: I know about book, gge, hugme, ilbot, json, november, nqp-rx, nqpbook, perl6-examples, proto, svg-matchdumper, svg-plot, temporal-flux-perl6syn, tufte, web
10:07 pmurias joined
pmurias what does plan * do? 10:07
10:07 mssm joined
moritz_ it's supposed to start planless testing 10:08
pmurias why do we need that?
moritz_ to distinguish it from a forgotten plan, I think 10:10
or from not running any tests at all, maybe
not really sure
pmurias it's not in all files 10:11
10:11 araujo joined
Trashlord hey 10:11
pmurias Trashlord: hi
moritz_: i think it would make sense to remove plan * 10:12
moritz_ pmurias: we should write a spec for testing
pmurias yes :) wanted to propose that to
moritz_ and I think it does make sense to distinguish loading Test.pm from starting testing 10:13
so *something* should stay that indicates starting of tests 10:14
pmurias the first call to a testing function could do that
moritz_ it could, but I'm not sure if I like that 10:15
pmurias we don't need to explicitly start testing
moritz_ aye; but that doesn't mean it's necessarily a good idea 10:16
pmurias if you don't have a plan or call done_testing the test harness will protest 10:17
moritz_ it might not make a difference for TAP, but maybe for other testing protocols
and while TAP is very good, it might not be the only thing worth considering (in the long range) 10:18
pmurias yes
but if you protocol needs an explicit start of testing the first call to the test function is enough
moritz_ unless there's a difference between running a test with no test cases, and running no test at all 10:19
suppose you have steps 1) load Test.pm 2) plan *; 3) run some tests; 4) done_testing 10:20
and you die silently before 3)
pmurias step 0 is start test
moritz_ how do you start test?
there might be difference if you die silently before running the first test
pmurias tests are one per file 10:21
moritz_ if you haven't declared that you want to run tests, some other protocols might not classify that as a failure
pmurias it would be the fault of the protocol 10:22
you can die between steps 1 and 2
moritz_ that would be the fault of the test writer :-)
anyway, I don't have a very strong opinion, but I'd like to hear what others have to say about it first 10:23
and somehow (probably irrationally) I like the step of explicitly saying "now we start to test"
pmurias i dispise such things
it seems like having to initialise some C library libtap_init(); 10:24
and Test::More doesn't require anything like that 10:25
10:26 Chillance joined
pmurias IMHO if we want something like that it should be run_test {ok 1;ok 2;...} 10:26
not plan *;ok 1,"first test";plan *;ok 1,"second test"; 10:27
moritz_ 'plan *' twice doesn't make much sense to me
pmurias calling 'plan *' at all doesn't make any sense to me 10:31
if we run a test we run it
10:32 jonasbn joined 10:33 payload1 joined
pmurias moritz_: or we can run multiple tests in the same program (then you need a way to distinguish when a tests starts and when it ends) 10:34
moritz_ if we want to test the language, we shouldn't rely too much on language features to work 10:36
so testing last, redo, next etc. inside a block (as introduced by run_test { ... }) might be dangerous
in general I like the idea, but not for the spectests 10:37
pmurias it would be more for doing fancy stuff like running multiple tests faster 10:38
moritz_ aye 10:39
pmurias we should likely ask the TAP guys about that once we have a solid spec of the basic stuff 10:40
moritz_ aye
I know that some of the TAP guys also have been working on subtests
pmurias that's why i suggested we ask them :) 10:41
moritz_ so at least a meta consens :-) 10:44
pmurias we could also ask them about plan * ;) 10:45
10:49 gfx left 11:08 orafu left, orafu joined
pmurias moritz_: are there any docs for Test.pm we could turn into a spec? 11:10
colomon ng: say -.2 ~~ -Inf..Inf 11:14
p6eval ng 1feef8: OUTPUT«0␤»
colomon ng: say -Inf before -.2 11:15
p6eval ng 1feef8: OUTPUT«0␤»
colomon ng: say -Inf < -.2 11:16
p6eval ng 1feef8: OUTPUT«1␤»
colomon ng: say (-Inf).WHAT
p6eval ng 1feef8: OUTPUT«Num()␤»
pmurias moritz_: my plan for getting writing a (semi-) spec for Test is to take the docs from ext/Test and the tests from t/02-test-pm correct and expand them and after everyone prune old stuff from rakudo's Test.pm
11:18 lest_away is now known as lestrrat, payload1 left
pugs_svn r29674 | colomon++ | [t/spec] Fix (some of the?) broken tests and refudge. 11:21
11:24 meppl joined, payload joined
colomon ng: my $a = "AAA"; say --$a; 11:26
p6eval ng 1feef8: OUTPUT«␤»
dalek kudo/ng: 7dc1b95 | (Solomon Foster)++ | t/spectest.data:
Turn on S02-builtin_data_types/range.t.
 11:27
11:38 masak joined
masak oh hai, #perl6 11:38
colomon o/
pmurias moritz_: hi 11:46
Trashlord hey 11:50
so when do you guys think perl 6 will be out? or do you not think about that? ;p
I mean, released officially
masak gee, the idea never occurred to us. :)
I guess we should set a release date or something.
Trashlord: how about Feb 18th? is that fine with you? 11:51
Trashlord I know a release date isn't planned
masak sure it is.
we have one every month.
we release Perl 6 regularly.
it's how good, functional open source is done. 11:52
"Release Early, Release Often." We do that.
Trashlord how is bad open source done?
masak in many other ways :)
Trashlord does Microsoft do bad open source? o.0 11:53
masak I'm not up-to-date with how much Microsoft does open source at all. not much, I suspect.
I know they have a couple of licenses registered with OSI, which is... bizarre. 11:54
pmurias some of the ghc guys are employed microsoft research
masak oh, that's nice.
yeah, they seem to get releases out, at least.
pmurias ghc is a good compiler 11:55
Trashlord was being sarcastic regarding Microsoft 11:56
masak The Net tends to treat sarcasm as damage and routes around it. 11:57
11:58 pmurias left 11:59 johnz left
Trashlord I'll make sure to memorize that 11:59
11:59 johnz joined, Lorn joined
masak Trashlord: anyway, when you were asking about a release date, what is it that you were hoping for in particular? a full-featured Perl 6 implementation? 12:00
12:00 bbkr left
masak any features in particular you would like to see? 12:00
Trashlord masak: yes. something like Perl 5?
-?
because I've played a bit with 6, and it looks promising, I enjoy it 12:01
masak cool.
I think so too.
dalek kudo/ng: 6a00fc0 | (Solomon Foster)++ | src/core/Range.pm:
Add Range.from and Range.to (why are these in the spec?!) and a very stupid but mostly functional implementation of Range.Num.
Trashlord I know that rakudo releases every month 12:02
I just installed it yesterday
tried some stuff 12:03
12:03 frettled sets mode: +o masak
frettled Try to hang on to it for more than an hour, eh? ;) 12:03
masak I promise :)
12:04 johnz left, johnz joined
jnthn lolhai 12:05
Trashlord lolhey
masak lolitsjnthn
jnthn :-)
frettled jnthnicious!
colomon jnthn: I've been busy slowing down make spectest 12:06
masak jnthnonderful!
colomon Files=115, Tests=3570
masak colomon++
jnthn colomon: Wow!! :-) 12:07
moritz_ rakudo: say 'foo'.WHAT
jnthn colomon: I gotta go debug a Java pile of hate for a bit again...will do some Rakudo hacking later on. :-) One thing from backlog (which I need more coffee to full read...)
moritz_ rakudo: say 'foo'.WICH
p6eval rakudo 1d4928: OUTPUT«Str()␤» 12:08
rakudo 1d4928: OUTPUT«Method 'WICH' not found for invocant of class 'Str'␤in Main (file src/gen_setting.pm, line 324)␤»
jnthn colomon: 1, 3, 5 ... *
moritz_ rakudo: say 'foo'.WHICH
p6eval rakudo 1d4928: OUTPUT«foo␤»
masak Trashlord: the thing with Rakudo that's been going on for several months now, by the way, is that we're rewriting the inner workings to better cope with future change. most development is taking place in a branch ('ng'), so there's not many changes between the last couple of releases.
moritz_ rakudo: say 23.WHICH
p6eval rakudo 1d4928: OUTPUT«23␤»
moritz_ rakudo: say 23.5.WHICH
p6eval rakudo 1d4928: OUTPUT«23.5␤»
moritz_ ng: say 23.5.WHICH
p6eval ng 7dc1b9: OUTPUT«47273655653440␤»
jnthn colomon: Is ... parsed interestingly?
colomon: That is, is it swallowing up the "1,3,5"?
Trashlord masak: right, so it's alright if I don't update every month?
12:08 ruoso joined
masak Trashlord: it's alright if you don't update for a year! just use whatever release fits your needs. :) 12:09
12:09 frettled sets mode: +o colomon
ruoso bom dia #perl6 12:09
Trashlord masak++
12:09 frettled sets mode: +o ruoso
colomon jnthn: as far as I can tell, if you do 1, 3, 5 ... *, it becomes +(1, 3, 5) ... * 12:09
masak ruoso: bom dia.
12:09 drbean joined
jnthn colomon: STD just says "list infix" 12:09
moritz_ good localtime() ruoso
frettled ruoso: bom bom bom, bom bom-badilla #perl6?
masak Trashlord: that said, we expect some release in Q2 (probably April) to really be something else.
jnthn colomon: aha 12:11
$P22 = "&infix:<,>"(1, 2, 3)
$P23 = "&infix:<...>"($P22, 5)
colomon: That's the code-gen for it.
colomon: So that's fine
12:11 cls_bsd left
jnthn ng: say Array.does(Positional) 12:11
p6eval ng 7dc1b9: OUTPUT«1␤»
jnthn ng: say Seq.does(Positional)
p6eval ng 7dc1b9: OUTPUT«1␤»
jnthn ng: say Parcel.does(Positional)
p6eval ng 7dc1b9: OUTPUT«0␤»
jnthn Ah.
colomon: And that's why it won't bind to @foo in the siggy. 12:12
colomon hmmm
Trashlord masak: awesome, looking forward to it 12:13
12:13 xomas joined, macae joined
colomon so, parcel should also be positional? 12:13
masak Trashlord: me too! I even tried to help the other day... :)
jnthn colomon: I'd have to check the spec. 12:14
colomon: My intuition is yes.
colomon ng: say (1, 2, 3) ~~ Parcel 12:15
p6eval ng 7dc1b9: OUTPUT«1␤»
colomon ng: say (1, 2, 3).Seq 12:16
p6eval ng 7dc1b9: OUTPUT«123␤»
jnthn ng: say (1,2,3).WHAT 12:17
p6eval ng 7dc1b9: OUTPUT«␤»
jnthn ...
From S02 12:18
Parcel Positional
So yes, it should.
colomon can I just copy the code from Seq?
it's one bit in the initialization, right?
(not logical bit, but bit of code.) 12:19
jnthn colomon: Not quite that simple, because at the moment Positional is much further into the bootstrap than Parcel, so I guess it'd be a case of trying to move it up first. 12:20
jnthn tries that to see if things explode. 12:21
colomon I think, now that I know what is going on, I can work around it for the moment.
or you can make thinks go boom, that's cool too. :)
12:22 clausi left
jnthn Hmm...well, it doesn't break the build...of course, I didn't make Parcel do it yet. 12:24
12:26 clausi joined
ruoso jnthn, colomon, but it's important to notice (I'm not really sure about it) that the positional access in the parcel is only in the first dimension 12:28
so if it has other parcels inside it you'll get them
colomon afk, no battery left
ruoso usually, the correct way to deal with Parcel is by using it as an Iterator
where you do specify the slice/flat context when getting the items 12:29
moritz_ is quited surprised by Neil Mowbray's error messages (on p6c)
jnthn ruoso: iiuc, binding it to @foo in a signature puts it in list context and it's no more a Parcel anyway. But yes, good to keep in mind.
colomon: Positional now does Parcel locally, will push. 12:31
12:32 cls_bsd joined 12:33 barney joined
jnthn moritz_: That error is just about certainly because it's failing to find the dynops 12:35
As for why, I'm less sure. :-/
moritz_ jnthn: yes, but why? Configure.pl checks if parrot is installed, including dev files
and usually one also gets an error message form failing to load such a thing, no? 12:36
jnthn moritz_: Annoyingly, Parrot seems to silently fail on some .loadlib directives. :-/ 12:37
moritz_: The problem is that it needs to find them at compile time, otherwise it doesn't know what in the PIR is an op or a function call. 12:38
ruoso jnthn, binding modifies the parcel? 12:39
jnthn ruoso: No, just enforces context on it 12:41
ruoso but it doesn't enforce flat context, does it? 12:42
moritz_ colomon: do you know if there was definitive decion on ?& et al (if it enforces context first)?
jnthn ruoso: Not sure what the latest spec is on that
ruoso: There were even commits in that area last night, that I didn't read yet... 12:43
ruoso iiuc it's the slurpy bind that forces flat
and the **@a bind forces slice
jnthn Still adapting to a post-@@ world. :-)
ruoso but default is non-modifying
binding doesn't enforce the contex
moritz_ isn't that ||@a now?
jnthn Those two are clear, but I'm not so sure on :(@foo) 12:44
ruoso moritz_, no... that's the other hand of **@a
jnthn :(@foo is rw) is clearly not going to do anything to it.
ruoso :(**@a) := (||@a)
moritz_ ruoso: ok, thanks
jnthn gotta go do @other-work-task for a while. 12:45
dalek kudo/ng: cb7f239 | jonathan++ | (2 files):
Parcel should do Positional.
 12:47
13:08 payload left 13:12 takadonet joined
takadonet morning all 13:12
13:17 SmokeMachine joined
masak takadonet: \o 13:18
right. I had 'a few' rakudobugs to report. 13:22
rakudo: (class foo {}).new
p6eval rakudo 1d4928: OUTPUT«Null PMC access in find_method('new')␤in Main (file <unknown>, line <unknown>)␤» 13:23
masak submits rakudobug
jnthn ng: (class foo {}).new
p6eval ng cb7f23: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤current instr.: 'perl6;Code;new' pc 11808 (src/builtins/Num.pir:27)␤»
jnthn Creative.
masak thanks.
rakudo: class A {}; class C {}; my A $a .= new; my C $c := $a; say $c
jnthn Should work though.
p6eval rakudo 1d4928: OUTPUT«A()<0x2b1564a60fb0>␤»
masak submits rakudobug
rakudo: use A::B 13:24
p6eval rakudo 1d4928: OUTPUT«Can't find ./A/B in @*INC␤in Main (file <unknown>, line <unknown>)␤»
jnthn masak: Think there already is one on binding not doing type checking.
masak I'm unsatisfied about the './'
jnthn That is a tad curious, yeah.
masak jnthn: ok. if you're sure that one's in RT, I won't submit it.
moritz_ I planned to report that for ages
masak rakudo: role A[B] {} 13:25
p6eval rakudo 1d4928: OUTPUT«set_pmc_keyed() not implemented in class 'Integer'␤in Main (file <unknown>, line <unknown>)␤»
masak submits rakudobug
moritz_ curious DB view: J."StartTime" AS "JStartTime", J."EndTime" AS "JStartTime",
masak here's one I'm not so sure about:
rakudo: subset A of Int; subset B of Num where { $^n !% 2 }; say A ~~ B 13:26
p6eval rakudo 1d4928: OUTPUT«Use of type object as value␤1␤»
masak I see why I get the warning, but should I?
also, should the answer be True here?
rakudo: subset B of Num where { $^n !% 2 }; say Int ~~ B 13:27
p6eval rakudo 1d4928: OUTPUT«Use of type object as value␤1␤»
masak that's simpler, I guess.
jnthn That should trivially be false in ng as Int !~~ Num now.
(OK, by spec it ain't)
However, I agree if it was, you should get the warning. 13:28
If not, you should get False.
And Int in general does not satisty the constraint.
13:28 eternaleye left
masak so nothing to see here, move along? 13:28
jnthn Pretty much, IMO.
masak ok.
finally, I have a spec question.
jnthn I'm not so fussed on making subsets behave specially on type objects.
masak S12:1482 says the following: 13:29
Int:_ where !*.defined
If `!*.defined` is allowed, whence the need for .notdef?
I know we talked about this before.
.notdef always seemed like a strange creature to me. I accept that it has a use in 'when' clauses.
jnthn std: :(Int:_ where !*.defined)
p6eval std 29674: OUTPUT«===SORRY!===␤Multiple prefix constraints not yet supported at /tmp/8ktmfDHjet line 1:␤------> :(Int:_ where !*.defined⏏)␤ expecting any of:␤ method arguments␤ type_constraint␤FAILED 00:01 110m␤»
masak but if !*.defined works, why do we need .notdef? 13:30
jnthn masak: I find it a tad surprising !*.defined should work in that context.
masak right.
jnthn Oh, hmm
Maybe not
Dunno. May be a relic of pre-.notdef days.
masak I'm pretty sure it is. 13:31
that part of the spec is older, at least.
jnthn One for TimToady++ to verify, I think.
masak nod.
rakudo: enum A <a b c>; enum B <a b c>; say &a.perl
p6eval rakudo 1d4928: OUTPUT«B::a␤»
masak submits rakudobug
jnthn ? 13:33
Heh 13:34
moritz_ it should be ambiguous
jnthn Oh
moritz_ I can never remember how to write that word :-)
jnthn Yeah, that.
jnthn dearly hopes somebody else takes on the enum implementation for ng.
mathw good afternoon #perl6 13:35
What's so horrible about enums? 13:36
jnthn mathw: I'm fed up of writing implementations of them.
mathw Should I put 'volunteering to look at the enum implementation in ng' before or after 'gouging my eyes out with a rusty nail' on my list of things I'd like to do today?
jnthn mathw: The recent spec changes might akshually make it better than the rusty nail. 13:37
mathw wow
I might have to have a look at those spec changes 13:38
jnthn ng: say 'enums' before 'rusty nail'
p6eval ng cb7f23: OUTPUT«1␤»
jnthn Well, there we go.
mathw lol 13:40
masak jnthn: I'd be happy to take on the enum implementation. I like the way it's spec'd nowadays. 13:41
it would be fun to make it happen.
mathw I'd like to try and follow along and learn something
masak cool. then I'll have someone to make rambling explanations to. :) 13:42
mathw :)
Just don't try them in Esperanto yet
I haven't learned any Esperanto since Monday
masak I promise.
mathw \o/ 13:44
jnthn masak++ mathw++ # enums!
mathw You're assuming my contribution will amount to anything more than pointless nitpicking 13:45
I shall try and be more useful than that
colomon moritz_: My understanding is that ?& coerces its arguments to Bool. I changed the implementation to work that way, too. :) 13:46
masak right. TimToady confirmed that, didn't he? 13:48
?& is a non-shortcircuiting version of &&
colomon masak: Yes.
masak: No.
ng: say 4 && 4
p6eval ng cb7f23: OUTPUT«sh: ./perl6: No such file or directory␤»
colomon > say 4 && 4 13:49
4
> say 4 ?& 4
1
And that 1 is actually Bool::True.
13:50 xomas left
colomon > (1, 3, 5 ... *).batch(10).perl.say 13:50
Method 'postcircumfix:<[ ]>' not found for invocant of class 'Parcel'
afk... desperately need more sleep 13:51
mathw 4 ?& 4 boils down to True AND True, I believe
masak colomon: oh, you're right. my statement is only true if I use the result of && in boolean context. 13:53
13:54 iblechbot left
masak four rakudobugs in one day. not bad, considering that master is not generating new bugs lately. :) 13:55
mathw masak: might also change depending on how your arguments coerce to bool, too
masak yeah. it's all very tricky, come to think of it.
14:05 xomas` joined 14:16 bbkr joined 14:24 slavik joined 14:29 andy1 joined, synth joined 14:31 pmurias joined
masak amazingly, the race is still on between GGE and ng. 14:34
to the extent that it was ever a race, of course. :)
14:35 drbean left
masak I have two features left until GGE is 100% PGE-compliant. 14:35
lookarounds and goals.
PerlJam masak: tortoise and hare? ;)
mathw impressive 14:36
masak PerlJam: maybe. I've managed to keep a fairly high activity wrt to GGE ever since August. but it does come and go.
I think it's my longest-running pet project ever, measured in the length of uninterrupted time I've managed to keep up an interest. 14:37
November might be slightly longer.
PerlJam masak: have you seen that code_swarm video I did of rakudo the other day? It's an interesting visualization of the "activity" of a project. 14:38
masak PerlJam: URL? 14:39
masak hasn't backlogged today, surprisingly
PerlJam feather.perl6.nl/~duff/rakudo-swarm.avi
masak looks
mathw masak: GGE is very impressive 14:42
I wish I had the time to try and understand how these engines work
masak is transfixed by that video
mathw I can comprehend a very basic regex engine, but anything high performance or with Perl-like features is just... beyond
masak mathw: thanks. stand by for my 7-part tutorial :)
mathw \o/
masak wow. 2008 was The Year It Happened.
PerlJam mathw: what perl-like features give you pause? 14:43
masak: wait, there's more :)
masak whoa!
2009! 14:44
PerlJam masak: if you want to play with your own code_swarm visualization, I put the config file I used in the rakudo repo in the tools dir.
mathw PerlJam: lookaheads, for one. I don't even understand how to use those...
masak I like the end especially. 14:45
pmichaud has like a big feaking pulsating aura around him, jnthn has a smaller one, and things just go kaboom, like.
it's like an action movie.
masak re-watches the end 14:46
it's at 3:42. November 15, 2009. probably ng being branched or something like that. 14:47
PerlJam yep
Trashlord heh
PerlJam (I used the ng branch for that visualization)
Trashlord I sometimes find myself at work, at 3:42
PerlJam master would have been much more boring towards the end 14:48
masak Trashlord: oh, I was unclear. :)
Trashlord: 3:42 in the vid.
Trashlord ha, alright then
Trashlord laughs
masak but I do believe pmichaud++ had some interesting sleep patterns while working on nqp-regex. 14:49
I had, too, during the more intense phases of GGE development.
mathw: re lookaheads, they're just a kind of subrule with a regex as an argument. and subrules are just regular methods. 14:50
mathw I will probably understand them at such time as I find myself actually needing to use one 14:56
masak what surprised me about them was that the 'foo' in <?before foo> isn't compiled at regex compile time, but at lookahead match time. 14:57
fortunately it's also cached.
mathw :) 14:59
good
masak but still, lookaraounds will incur a runtime delay while compiling those 'foo'-like things. that ought to make them significantly slower than other regexes, since compilation is heavy.
mathw yeah so I'm probably better off avoiding them in the general case
except where it makes something that would otherwise take a load of clunky code much more elegant
masak dunno. they're definitely useful.
mathw and sometimes ?before does seem tempting, if I really understood what it does :) 15:00
masak it's a zero-width match at the current location. 15:01
'zero-width' means that even if the thing you're matching has some length, your current position doesn't move.
jnthn masak: Basically, it says "does the next thing match this pattern", but without actually capturing or consuming any characters.
erm 15:02
tht was for mathw
masak jnthn: :)
jnthn :-)
mathw yay
ma<tab> strikes again
mathw gives jnthn a beer
jnthn \o/
jnthn has also managed to find the annoying bug in the Java thingy he's hacking on, after many hours searching.
mathw yay
jnthn That greatly increases the chance of me getting Rakudo time today. :-) 15:03
mathw yay xx (99..999).pick 15:04
jnthn I have to say that this code-base gave me a new respect for TimToady's wanting making classes final etc a program-level decision rather than one for the class though.
*loads* of stuff has been marked final, I guess for performance win.
ut it
But it means I can't just subclass the thingy that is doing the Wrong Thing to fix it, I gotta patch the actual library. 15:05
PerlJam That TimToady guy is one smart cookie.
;)
jnthn Yeah, but tbh until now, I'd not really grasped how annoying "final" decided by a class could be. :-) 15:06
mathw I had
I've been bitten by that
What with being a C++ programmer
jnthn You're suddenly making my need to do Java feel less bad. ;-)
mathw And having to deal with the need for polymorphic behaviour, but the base class didn't declare any of its methods virtual
so polymorphic behaviour DO NOT HAVE
jnthn eww
mathw quite
jnthn Bad choice of default, imo. 15:07
mathw now, virtual methods do impose a runtime performance hit
but it's not a very big one these days
C++ is a powerful language indeed, and I'd still rather use it than Java
but it's got some crazy shit in it
PerlJam munches on coworker's home-made pork jerky $coworker++
mathw The type system is still crippled by its need to be C-like
and C++0x is taking entirely too long 15:08
PerlJam mathw: what would you rather C++ than Java exactly?
s/what/why/
jnthn EWW jerky :-O 15:09
mathw because C++ lets me write types which behave indistinguishably from built-in types in all important respects
I can have a bignum type that reacts to arithmetic operators in precisely the expected manner
masak mathw: could you tell me a bit about X++0x? your 'taking entirely too long' comment piqued my interest. 15:10
er, s/X/C/
15:10 nacho joined
mathw I can write classes which do hugely crazy memory-sharing tricks internally and not care when I'm using them because it just happens behind the scenes 15:10
masak: the standardisation committee is working very slowly, basically 15:11
There's some good stuff in the drafts
PerlJam mathw: I haven't used Java much, but I couldn't imagine someone *preferring* C++ :-) Thanks.
masak mathw: tell me about the good stuff.
there's memory management, no? 15:12
mathw masak: unfortunately they dropped template concept checking, which looked like a great idea
masak why'd they drop it?
mathw unproven
they decided that something developed for spec rather than developed in real use was too risky
they've been burned by things like that before it seems
masak ah. makes sense.
it -is- very useful to have at least one working implementation of the spec'd thing. 15:13
cf Pugs, Rakudo.
mathw uniform initialisation is an interesting one
masak whuzzat?
mathw gives us a object construction syntax that's the same for absolutely everything
and which lets you write a handler for it for a user-defined type 15:14
Trashlord masak: you have my highest respect
mathw so that standard containers will have a literal initialisation syntax
Trashlord I must leave now, I have to create music, so see you guys later
15:14 Trashlord left
masak Trashlord: thank you, for whatever reason. 15:14
dang, missed him :)
15:14 cognominal left
mathw the other thing I usually think of as exciting is the tremendously mundane rvalue references 15:15
which should lead to some performance gains
masak 'tremenduously mundane'?
mathw yeah it's not particularly flashy
masak that can indeed be good.
mathw but it means you'll be able to return temporary objects from functions by reference instead of by value
pmichaud good morning, #perl6
masak has my S26 email arrived yet? I think S26 suffers from being overly flashy. 15:16
mathw ordinary references couldn't handle that, because the object would've been destroyed already
masak pmichaud: \o
jnthn pmichaud: Morning
mathw rvalue references enable the concept of a 'move constructor' as well, which is a kind of destructive copy of an object which is going out of scope. They're basically all about cutting out unnecessary copying and object creation.
15:17 SmokeMachine left
masak mathw: that sounds nice. 15:18
mathw yeah
masak "The Art of the Metaobject Protocol" starts out by saying that compilers should be able to optimize for various object creation/destruction scenarios, but they generally don't have enough information to do that at compile-time. 15:19
mathw there's library stuff to go with it
15:19 cognominal joined
mathw but yeah, overall I do like C++ despite its flaws 15:19
maybe just because I've spent so much time with it, it's like a comfortable jumper that's developed a hole
The only thing that really grates me regularly is that it has at the language level utterly zero comprehension of threading 15:21
15:21 nacho left
mathw Oh and no anonymous functions - but C++0x adds those too, which is exciting, even if they're not as flashy as Apple's 'blocks' extension which are full-blown closures. 15:21
it will make using STL algorithms much nicer :) 15:22
15:22 am0c joined
moritz_ whenver I go back to a language that has no closures, I feel so lost 15:22
15:23 Guest55887 left
PerlJam mathw: how do you pronounce C++0x? :-) 15:23
jnthn moritz_: When C# 3.0 added such things, it felt like a whole different langauge to me.
mathw yeah I love closures
My C++ would benefit greatly from them 15:24
jnthn Programming in C# was suddenly funner.
mathw and I also think they'd make it a lot easier to write good multithreaded apps
even without any extra multithreading support
although I think maybe C++0x also has a standardised syntax for TLS
can't remember
anyway, closures rock
PerlJam: "see plus plus oh eks"
although it should be called C++1x now, because it's definitely not coming out before 2010... 15:25
15:25 cjk101010 left
mathw and then we have to wait for compilers to propogate 15:25
so I'm not going to get to use this stuff at work for ages
seeing as how our compilers don't even ship TR1
I also need to get to the bottom of why one of the senior members of my team thinks smart pointers are the source of all evil 15:26
PerlJam smart pointers use reference counting?!? 15:29
weird
mathw some of them do
std::auto_ptr<> doesn't, but you can only have one of those pointing to a given object at any given time or Weird Things Happen. 15:30
(i.e. it tends to get destroyed before you expect it, and your program explodes) 15:31
PerlJam one of these days I'll have to give C++ a try again (I haven't touched C++ in over a decade) 15:32
mathw Well I'm firmly of the opinion that exposure to more languages is of great benefit to any programmer 15:33
even Java can teach you something
masak full ack.
jnthn Agree.
PerlJam yep. And C++ looks like a different animal than the one I knew, so it'll be a brand new language to me :)
mathw it's changed since I first learned it 15:34
jnthn wouldn't mind giving F# a look
mathw A lot of the changes have to do with people figuring out terrifyingly complicated things to do with the template system
F# interests me, as a Haskell nut
jnthn It interets me as a guy who once did ML.
mathw I need a little project to write in it
jnthn :-)
moritz_ Fortress also looks like a cool language for some purposes
mathw Assuming I can get around the ML syntax 15:35
which I don't know very well
I've read plenty of ML, but I always wrote Haskell
moritz_ I found that reading a language doesn't really teach you its syntax
mathw I took a look at Clojure earlier, after masak's tweet, but I came to the same conclusion I did last time I looked at Scheme
I simply can't handle Lisp syntax
PerlJam heh
mathw it makes my eyes go funny 15:36
15:36 SmokeMachine joined
mathw I just don't have the inclination to spend the time required to get used to it 15:36
15:36 xomas` left
mathw I've not heard of Fortress 15:36
15:36 xomas_ joined, xomas_ left, xomas_ joined
moritz_ it's a Java redesign of Fortran 15:36
that sounds so negative 15:37
mathw yes
moritz_ s/Java/Sun/
masak Fortress shares a lot of things with Perl 6.
mathw you've just mentioned two of my least favourite languages together
masak it's meant to be growable.
moritz_ it's a new language from Sun that tries to be *very* good for numerical calculations
mathw looks it up
masak it does a lot of things with definable operators.
15:38 dugg joined
moritz_ en.wikipedia.org/wiki/Fortress_(pro..._language) 15:38
masak research.sun.com/projects/plrg/Publ...ss.1.0.pdf
PerlJam Another Guy Steele language.
moritz_ "Fortress is also designed to be both highly parallel and have rich functionality contained within libraries, drawing from Java but taken to a higher degree. For example, the 'for' loop is a parallel operation, which will not always iterate in a strictly linear manner depending on the underlying software and hardware. However, the 'for' loop is a library function and can be replaced by another 'for' loop of the programmer's liking rather than being built into th
... rather than being built into the language." 15:39
that sounds a wee bit like Perl 6 :-)
xinming PerlJam: What does "Guy Steele" mean please?
moritz_ en.wikipedia.org/wiki/Guy_L._Steele_Jr. 15:40
PerlJam xinming: He's a big Lisp-head at Sun who has influenced several different technologies
xinming Dman, It's a person name, I thought Guy Steele is kind of style. :-P
PerlJam: Thanks.
masak I can heartily recomment Steele's "Growing a Language". one of the best talks I've ever read.
gtg & 15:41
15:41 masak left
PerlJam I always think of Guy Steele as being old, so I'm always surprised at how young he actually is. 15:43
maybe it's just becuase of how much older than me he is that I think of him as "old" 15:44
15:44 nihiliad joined
mathw Well it's good really, because it means he'll be around to design more cool stuff for a while yet 15:44
moritz_ the more you read about how much somebody has done, the more you think of him as being "old" 15:45
PerlJam moritz_: there's that too
mathw yeah but then when you find out how young they are, you end up feeling inadequate
moritz_ if you chose to compare yourself to the top 0.01% of programmers you likely feel inadequate 15:46
mathw yes
I try not to do that very often
15:46 clausi left, cl58 joined
mathw I prefer to only compare myself to people in fields I'm actually good at :) 15:47
PerlJam How do I know if the people I compare myself to are in the top 0.01% ? :)
15:47 cl58 left
moritz_ my mother happens to be a psychologist. She always told that the only person I should compare myself to is me 15:47
it took me some years to realize, but she is right
xinming I don't prefer to anyone as I always found frustration...
PerlJam moritz_: more proof that you should always listen to your mother. 15:48
mathw We should all listen to moritz's mother
xinming At last, I realized, I shouldn't compare with others, Otherwise, I'll live in sorrow. :-)
15:49 justatheory joined
Su-Shee or, you're just not male and consider this competition stuff entirely different anyways. ;) 15:49
moritz_ Su-Shee: maybe you listened to your mom :-) 15:50
PerlJam Su-Shee: comparison != competition :) 15:51
Su-Shee no, my learned values are plainly different from yours. I had to actually learn how to deal with all the male measurements of competition.
PerlJam: as you (meta-you in the meaning of "you guys coding stuff) practice it, it's usally "compare to compete".
xinming moritz_: But there is no much luck for Chinese people actually, parents always try to compare you with other people, and say, Look, How good they are, You should follow them up.
15:54 justatheory left
xinming goes to sleep. bye all. :-) 15:54
PerlJam goodnight xinming
colomon mathw: you think you've got it bad waiting for C++0x -- I had people I work with git upset last fall when I suggested I might not support Visual C++ 6.0 bugs anymore in my code.... 15:56
mathw colomon: yeah that's pretty bad, but we do use Visual C++ 2005 at work 15:57
colomon mathw: I use 2005 too (mostly, I do have 2008). 6.0 is like 1997 or 1998. 15:58
15:59 Kvorg joined, Patterner left
colomon mathw: my point was just if the trend continues, I don't get to use C++0x for production code until 2025 or so... 16:01
16:01 Psyche^ joined, Psyche^ is now known as Patterner
colomon and by then we'll have working Perl 6 to .NET compilers, so it will be pointless anyway. :) 16:01
16:01 nacho joined 16:02 jonasbn left
PerlJam Hmm. 16:03
jnthn Perl 6 on .Net would be cool. :-)
PerlJam listens to Growing a Language (keynote OOPSLA 1998)
moritz_ started too 16:04
pmurias moritz_: what's wrong with comparing yourself to others? 16:05
PerlJam moritz_: about 18 minutes in is an interesting bit that made me think of perl.
moritz_ pmurias: you can only compare comparable things 16:06
pmurias: and depends on what you compare, you might not be comparable to other people
16:06 xomas_ is now known as xomas
moritz_ pmurias: and you might not notice that, which results in unhappiness 16:06
PerlJam I wonder how we can get people like Guy Steele to help out with Perl 6 (and if that would even be beneficial. It seems so, but what do I know?)
16:10 cjk101010 joined 16:22 uniejo left 16:24 xomas left 16:34 k23z__ joined
arnsholt is contemplating implementing a rudimentary gather/take in ruby 16:36
Tene arnsholt: Cardinal, the ruby-on-parrot project, has it. :) 16:54
16:55 macae left
arnsholt Cool. I'm working on an existing project (which uses Rails) so I suspect ruby-on-parrot might not be an option 16:56
But I've found a module that supplies dynamic scope, so it should be reasonably simple to implement
17:00 uniejo joined 17:02 rindolf joined
rindolf Hi all. 17:02
OK, I think I'll learn some Perl 6 now.
Tene hi
moritz_ rindolf: have fun 17:03
rindolf moritz_: thanks.
Hi Tene
Tene Nice to see you here, rindolf.
rindolf Tene: :-) 17:04
sjohnson Perl 6 will bring you happiness of the heart
diakopter the hope of which, anyway. 17:06
sjohnson one of my favourite features is the: if (5 > y > 10) stuff 17:07
err
rindolf sjohnson: that is always false.
sjohnson 5 < y < 10
i mean
colomon ng: my $a = NaN; say 5 > $a > 10
p6eval ng cb7f23: OUTPUT«0␤»
moritz_ rindolf: unless y returns something of a type with a weird overloaded infix:«>» operator 17:08
rindolf moritz_: heh, heh.
moritz_ I've learned to be careful with words like "all" around here
rindolf How long is {{ say [+] (1..999).grep( { $_ % 3 == 0 || $_ % 5 == 0 } ); }} supposed to take? 17:09
colomon huh. now you've got me wondering if Rat > Rat works "correctly".
moritz_ rindolf: as short as possible 17:10
rindolf moritz_: :-)
moritz_ takes 5s on my machine
17:10 cdarroch joined, cdarroch left, cdarroch joined
pugs_svn r29675 | lwall++ | [Spec] squash [;] fossils noticed by eternaleye++ 17:11
jnthn 5s?!
17:11 cotto_working left
moritz_ yes 17:11
diakopter rakudo: say [+] (1..999).grep( { $_ % 3 == 0 || $_ % 5 == 0 } );
p6eval rakudo 1d4928: ( no output )
colomon btw, jnthn, getting [+] working again in ng is something I'm interested in tackling soon. 17:12
diakopter timed out on that vps.
jnthn colomon: Cool
diakopter rakudo: say [+] (1..99).grep( { $_ % 3 == 0 || $_ % 5 == 0 } );
p6eval rakudo 1d4928: OUTPUT«2318␤»
diakopter rakudo: say [+] (1..199).grep( { $_ % 3 == 0 || $_ % 5 == 0 } );
p6eval rakudo 1d4928: OUTPUT«9168␤»
colomon jnthn: I'm pretty sure I'm going to need help on the grammar for that. :) 17:13
TimToady rakudo: say [+] (1..999).grep: * !% 3|5
p6eval rakudo 1d4928: OUTPUT«any([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
TimToady rakudo: say [+] (1..999).grep(* !% 3|5)
p6eval rakudo 1d4928: OUTPUT«any([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
diakopter rakudo: say [+] ((1..999).grep: * !% 3|5)
p6eval rakudo 1d4928: OUTPUT«any([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
TimToady rakudo: say [+] (1..999).grep(* !% 3&5) 17:14
p6eval rakudo 1d4928: OUTPUT«all([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
TimToady hmm
diakopter rakudo: say [+]((1..999).grep: * !% 3|5)
p6eval rakudo 1d4928: OUTPUT«any([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
diakopter rakudo: say [+](1..999).grep: * !% 3&5 17:15
p6eval rakudo 1d4928: OUTPUT«all([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
TimToady grep doesn't appear to be collapsing the junction
jnthn rakudo: say 15 !% 3&5
p6eval rakudo 1d4928: OUTPUT«all(Bool::True, 5)␤»
diakopter rakudo: say [+](1..999).grep: * !% (3&5)
p6eval rakudo 1d4928: OUTPUT«all([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
diakopter rakudo: say 15 !% (3&5)
p6eval rakudo 1d4928: OUTPUT«all(Bool::True)␤»
diakopter base case 17:16
jnthn Precedence.
TimToady yes, the ! should be collapsing it
jnthn rakudo: say 15 !% 3&5; say 15 !% (3&5);
TimToady hmm
17:17 justatheory joined
p6eval rakudo 1d4928: OUTPUT«all(Bool::True, 5)␤all(Bool::True)␤» 17:17
TimToady right
rakudo: say [+] (1..999).grep(* !% (3|5))
p6eval rakudo 1d4928: OUTPUT«any([3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99, 102, 105, 108, 111, 114, 117, 120, 123, 126, 129, 132, 135, 138, 141, 144, 147, 150, 153, 156, 159, 162, 165, 168, 171, 174, 177, 180, 183,
..186, 189…
jnthn (Dunno if Rakudo's correct there.)
It's still failing to collapse the junction to a single boolean value though.
TimToady there's something that ought to be collapsing that isn't
17:17 cotto_working joined
jnthn say [+] (1..999).grep((* !% (3|5)).Bool) 17:18
rakudo: say [+] (1..999).grep({(%_ !% (3|5)).Bool})
p6eval rakudo 1d4928: OUTPUT«Method 'Bool' not found for invocant of class ''␤in Main (file src/gen_setting.pm, line 324)␤»
jnthn rakudo: say [+] (1..999).grep({?(%_ !% (3|5))})
p6eval rakudo 1d4928: ( no output ) 17:19
jnthn :-/
TimToady my feeling is that * !% junction should just be returning true or false, since ! is boolean
jnthn That'd make sense to me too.
TimToady and, in fact, the ! metaop is really outside in the same sense as not() 17:20
17:20 rv2733 left
TimToady that would be a useful test for someone to write 17:20
17:23 orafu left
TimToady since, in any case $a != (2|3) should probably really mean not($a == (2|3)) if we following English semantics 17:23
17:23 orafu joined
TimToady assuming we rewrite != to ![==] and always hoist the negation to outside so that junctions never see negative operators 17:25
this would provide Least Surprise 17:26
rindolf How do I prepend a string to $s with p6? 17:29
pmichaud $s ~= 'string'
oh, prepend
$s = 'string' ~ $s
or 17:30
$s = "string$s"
rindolf pmichaud: ah.
TimToady hmm
$s [R~]= 'string' 17:31
pmichaud I thought about that briefly.
PerlJam TimToady++ you beat me to it
TimToady but might not actually work
pmichaud afk, lunchish
rindolf my $s = "Sophie"; $s R~= "Hello "; say $s doesn't work. 17:32
TimToady probably parses as R[~=]
pmichaud I'm certain it doesn't work in Rakudo yet.
rindolf Confused at line 1, near "[R~]= \"Hel"
pmichaud Rakudo doesn't understand multiple metas.
TimToady still writes out all the cases on spec?
moritz_ std: 1 [R~]= 2
p6eval std 29675: OUTPUT«ok 00:01 107m␤» 17:33
pmichaud ng will handle metas (or at least attempt it... not sure about multiple metas yet :-)
TimToady ng: my $s = "Sophie"; $s [R~]= "Hello "; say $s 17:34
p6eval ng cb7f23: OUTPUT«Confused at line 1, near "$s [R~]= \""␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
pmichaud "will" === "not yet" :-|
17:34 cognominal left
TimToady ng: my $s = "Sophie"; "Hello " R~= $s; say $s 17:35
p6eval ng cb7f23: OUTPUT«Confused at line 1, near "\"Hello \" R"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
17:37 macae joined
TimToady perl6: my $s = "Sophie"; substr($s,0,0) = "Hello "; say $s 17:37
p6eval rakudo 1d4928: OUTPUT«Sophie␤»
..pugs: OUTPUT«Hello Sophie␤»
..elf 29675: OUTPUT«Can't modify non-lvalue subroutine call in scalar assignment at (eval 125) line 4, near ""Hello\ ")"␤ at ./elf_h line 5881␤»
TimToady er, what did rakudo do with the Hello? 17:38
ng: my $s = "Sophie"; substr($s,0,0) = "Hello "; say $s
p6eval ng cb7f23: OUTPUT«Cannot assign to readonly value␤current instr.: '&infix:<=>' pc 15051 (src/builtins/Junction.pir:113)␤»
pmichaud substr() returned a value
and the assignment overwrote it
[particle] IT EATED IT
pmichaud because rakudo (master) allows writing over "constant" values 17:39
TimToady perl6: my $s = "Sophie"; substr($s,0,0, "Hello "); say $s
p6eval elf 29675, pugs: OUTPUT«Sophie␤»
..rakudo 1d4928: OUTPUT«too many positional arguments: 4 passed, 4 expected␤in Main (file <unknown>, line <unknown>)␤»
pmichaud or, phrased differently, rakudo's substr doesn't return a modifiable lvalue for replacement
TimToady should probably split out an insert() instead of overloading a 4-arg substr 17:40
perl6: my $s = "Sophie"; $s ~~ s[^] = "Hello "; say $s 17:41
p6eval rakudo 1d4928: OUTPUT«s/// not implemented, try .subst as workaround at line 10, near " = \"Hello "␤in Main (file <unknown>, line <unknown>)␤»
..pugs: OUTPUT«Error eval perl5: "if (!$INC{'Pugs/Runtime/Match/HsBridge.pm'}) {␤ unshift @INC, '/home/p6eval/.cabal/share/Pugs-6.2.13.14/blib6/pugs/perl5/lib';␤ eval q[require 'Pugs/Runtime/Match/HsBridge.pm'] or die $@;␤}␤'Pugs::Runtime::Match::HsBridge'␤"␤*** '<HANDLE>' trapped by operat…
..elf 29675: OUTPUT«Parse error in: /tmp/6C27J5a5iZ␤panic at line 1 column 18 (pos 18): Can't understand next input--giving up␤WHERE: my $s = "Sophie"; $s ~~ s[^] = "Hello "; say $s␤WHERE: /\<-- HERE␤ STD_red/prelude.rb:99:in `panic'␤ STD_red/std.rb:76:in `scan_unitstopper'␤
..STD_re…
rindolf rakudo: say ((5,6,3) X+ (100,200,300)).perl
p6eval rakudo 1d4928: OUTPUT«[105, 205, 305, 106, 206, 306, 103, 203, 303]␤»
rindolf say ((5,6,3) X(min) (100,200,300)).perl ==> how do I write it properly? 17:43
TimToady rakudo: say (5,6,3 X[min] 100,200,300).perl
p6eval rakudo 1d4928: OUTPUT«[5, 100, 6, 100, 3, 100]␤»
TimToady rakudo: say (5,6,3 Xmin 100,200,300).perl 17:44
p6eval rakudo 1d4928: OUTPUT«[5, 5, 5, 6, 6, 6, 3, 3, 3]␤»
moritz_ std: 1 Xmin 2
p6eval std 29675: OUTPUT«ok 00:01 105m␤»
TimToady rakudo doesn't understand [] infixes
17:44 cotto_working left
TimToady so misparses X[min] as X [min] 17:44
17:45 cotto_working joined
TimToady rakudo: say 1 [+] 2 17:45
p6eval rakudo 1d4928: OUTPUT«Confused at line 10, near "[+] 2"␤in Main (file <unknown>, line <unknown>)␤»
TimToady ng: say 1 [+] 2
p6eval ng cb7f23: OUTPUT«Confused at line 1, near "say 1 [+] "␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
TimToady std: say 1 [+] 2 17:46
p6eval std 29675: OUTPUT«ok 00:01 106m␤»
TimToady std: say 1 [[[[[+]]]]] 2
p6eval std 29675: OUTPUT«ok 00:01 106m␤»
TimToady S04:4394 17:47
er, S03:4394
17:48 cognominal joined 17:51 jaldhar left, jaldhar joined
rindolf say (^10).list.perl 17:56
colomon ng: say (^10).list.perl 17:57
p6eval ng cb7f23: OUTPUT«0..^10␤»
TimToady ng: say (^10).flat.perl 17:59
p6eval ng cb7f23: OUTPUT«Method 'flat' not found for invocant of class 'Range'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady ng: say (^10,).perl
p6eval ng cb7f23: OUTPUT«(0..^10, )␤»
17:59 dakkar left
TimToady well, flat should do it 17:59
colomon ng: say (^10).batch(10).perl
p6eval ng cb7f23: OUTPUT«Method 'batch' not found for invocant of class 'Range'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon ng: say (^10).iterator.batch(10).perl 18:00
p6eval ng cb7f23: OUTPUT«(0, 1, 2, 3, 4, 5, 6, 7, 8, 9)␤»
TimToady Range probably needs an Any/Cool loopback to listify
or flatten, maybe 18:01
colomon ng: my @a = 1, 2; say &infix:<cmp>(|@a);
p6eval ng cb7f23: OUTPUT«Confused at line 1, near "say &infix"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
colomon TimToady: or possible Iterable needs some additional functions. It's all still kind of sketchy at the moment. :)
ng: my @a = 1, 2; sub f($a, $b) { say "a = $a and b = $b"; }; f(|@a); 18:03
p6eval ng cb7f23: OUTPUT«Confused at line 1, near "f(|@a);"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
TimToady this whole loopback scheme is all about tagmemics; if you use X as if it were a Y, then there's a semantic mechanism to do that.
verbing nouns, and such
a lot of methods use their invocant as a list or slice 18:04
so the loopback gives the general mechanism so we *don't* end up scattering the coercions all over the palce
*place
colomon TimToady: sure, but Range actually is Iterable, which should mean it knows how to be used as a list or slice.
(I would think, anyway.) 18:05
TimToady but it's the Coolish method that figures that out
Oh, I got something that doesn't respond to .batch, but I know that it's Iterable, so I'll iterate it for .batch
likewise for grep etc 18:06
we shouldn't be putting a .grep method in everything iterable
DRY, and all that 18:08
PerlJam what is .batch again?
TimToady N at a time
PerlJam gotcha 18:09
TimToady .batch(10) asks for 10, while .batch() just asks for "some"
colomon TimToady: right, but if the .grep method was on Iterable itself, you wouldn't be repeating yourself for each iterable class.
PerlJam er, .batch() asks for "some"? That makes it behave like a lazy list? 18:10
colomon mind you, I don't think .grep is the right granularity, it should be something like .list.
PerlJam: even if you ask for 10, it might still be lazy.
moritz_ PerlJam: laziness isn't guarantueed like in Haskell
PerlJam colomon: aye, it's not the laziness I was worried about, but the listiness. 18:11
(dunno why I typed "lazy list" then, but weird things happen)
colomon PerlJam: I'm not sure there's a distinction between the two anymore?
TimToady anything .batch hands back is already evaluated
it's the "how eager I want to be right now" method 18:12
18:12 ShaneC1 joined
colomon ng: say (^10).grep{{$^a %3 == 1}).batch(10).perl.say 18:12
p6eval ng cb7f23: OUTPUT«Confused at line 1, near "say (^10)."␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
TimToady in supply and demand, it's the demand
colomon ng: say (^10).grep({$^a %3 == 1}).batch(10).perl.say
p6eval ng cb7f23: OUTPUT«(1, 4, 7)␤1␤»
PerlJam okay, the jibes with what I was thinking even if I wasn't expressing it clearly. 18:13
colomon TimToady: so in fact, .grep already works this way. It calls .list internally. :)
TimToady well, should probably be calling flat now...
but yeah, Iterable is a good place for that 18:14
it's the non-Iterable that needs an Any entry
to just return itself as a singleton iterator
so that 42.grep works
anyway, .list doesn't really mean much now, and it certainly doesn't flatten 18:16
18:17 Su-Shee left
TimToady nowadays .list means only "get the iterator from this and return it in a Parcel for later contextualization" 18:22
and .slice and .flat imply .list 18:23
colomon Does anyone out there have a clue how one would go about implementing f(|@a)? 18:24
I think I've patched it in the grammar, but that's only the first step...
(in ng, I mean.) 18:25
jnthn Can maybe get away with something like what master does for now. 18:26
But really probably needs constructing a capture.
If latest spec still says that's what it does. :-)
colomon jnthn: any idea where that would be located in master? 18:28
(I'm willing to accept half-measures if they work. :)
jnthn colomon: In the old actions.pm, I think. 18:31
colomon: It probably examines the sigil of the variable that follows and sets .flatten() on the past node, or maybe .flatten() and .named() 18:32
colomon named only if the sigil is % 18:33
18:34 hercynium left
colomon You think that will work in ng? 18:34
colomon is in almost but not quite completely, over his head. 18:35
sjohnson hi colomon 18:36
colomon sjohnson: o/
18:36 stephenlb joined 18:37 stephenlb left
jnthn colomon: I don't see why not. It'll have the same imperfections master has I guess, that's all. 18:37
hi sjohnson
sjohnson jnthn: hi! may I ask you an OT question
18:37 stephenlb joined 18:38 Su-Shee joined
sjohnson just wondering if you are like me, in that you use Windows for desktop use, yet use things like Strawberry Perl, unxtools on sourceforge, and dislike cygwin 18:38
jnthn I tried cygwin and never got on with it.
sjohnson unxtools == sourceforge.net/projects/unxutils/
jnthn I don't use that specifically, I did grab some utilities from some other gnu tools porting project though. I forget where. :-) 18:39
I've tended to use ActivePerl a bit more, though mostly because I just had it around for longer. 18:40
I've installed Strawberry elsewhere too though.
sjohnson i am finding i like Strawberry a bit more 18:41
as it feels less corporate, the install is simpler and less nonsense, and for me, CPAN works a bit better
i've tried ubuntu and others about half a dozen times, and i find that Microsoft's OS's (the good ones) aren't really all that bad 18:42
jnthn Yeah, I found CPAN smoother on Strawberry.
colomon jnthn: was it DEBUG that did say from NQP?
jnthn iirc, yes 18:43
sjohnson: Oh, I find XP plenty usable, and 7 seems decent enough.
The trick is to try and skip the sucky releases. :)
sjohnson jnthn: yes i have found i prefer 7 more than XP, but XP was very good. I strongly disliked Vista though, as i'm sure that isn't a surprise to anyone 18:44
Vista is like the Windows ME of >2000 OS's
i'm almost.. considering... *gasp*.. paying for windows 7 for my home use
as the prices are very reasonable now. 100+ bucks or so, instead of 500 like the days of yore 18:45
with XP
18:45 cjk101010 left 18:46 justatheory left
colomon oh, that compiled. 18:48
I'm kind of scared now.
"flattened parameters must be a hash or array" 18:49
jnthn :-) 18:50
Wonder whoz op with that.
18:50 justatheory joined
colomon surely that should be positional or associative now, right? :) 18:50
jnthn I thought it looked at the sigil 18:51
In which case, it's just that it's looking at it the wrong way. :-)
I want to go find some nom now, but I'll not be gone long...feel free to nopaste the diff and I'll look in a moment when I'm back.
lisppaste3 colomon pasted "experimental code to implement f(!@a) in ng" at paste.lisp.org/display/94713
colomon that error message isn't actually from rakudo -- must be in parrot? 18:53
lisppaste3 colomon annotated #94713 "new code in context" at paste.lisp.org/display/94713#1 18:55
19:00 uniejo left 19:13 xomas_ joined, xomas_ left, xomas_ joined 19:14 uniejo joined 19:18 cl58 joined
jnthn back 19:24
colomon: I thought about prefix:<|> while away.
colomon: I don't think porting the old impl from master will fly :-(
colomon :(
jnthn colomon: Because of (a) our arrays has-a ResizablePMCArray now rather than is one, but also (b) it'd probably play badly with laziness. 19:25
So I figure we're gonna have to do it "properly"
Whatever properly means.
colomon btw, I have .reduce working on binary functions now. (easiest workaround for lack of f(|@a))
jnthn OK, cool.
colomon how would it play badly with laziness? 19:26
oh, i think i see
jnthn Parrot would want to have all of the elements, I suspect. 19:27
colomon yeah
jnthn But (a) is a far bigger problem.
colomon it doesn't know about laziness at all.
jnthn (In terms of taking the approach master does, at any rate)
Right
I'll think about it, I don't have an immediate good answer.
colomon okay
I'm going to work on cleaning what I've got that works and pushing it. 19:28
jnthn Cool 19:31
I noticed you won us back a good bunch of tests too :-) 19:32
colomon: Is there anything I should look at that's blocking you?
(Aside from prefix:<|> but I think that needs a bit more thought...) 19:33
19:33 old_gray_bear joined
colomon jnthn: as far as I know most Rat fixes should be straightforward with multi dispatch working. 19:34
there is the issue with weirdness with in getting the right $_ in grep. 19:35
ng: (^10).grep({say $_; 1;}).eager 19:36
p6eval ng cb7f23: OUTPUT«Mu()␤Mu()␤Mu()␤Mu()␤Mu()␤Mu()␤Mu()␤Mu()␤Mu()␤Mu()␤»
19:36 mberends joined
colomon that's not exactly blocking, but we'd win back a fair number of tests it it worked. 19:36
I suspect that one is either really easy or really hard 19:37
pmichaud back 19:40
dalek kudo/ng: 2a0400a | (Solomon Foster)++ | src/core/Any-list.pm:
Add Any.reduce. Only works on binary functions for the moment.
 19:41
kudo/ng: 9c17cb6 | (Solomon Foster)++ | src/Perl6/Grammar.pm:
Add prefix:sym<|> to the grammar.
pugs_svn r29676 | colomon++ | [t/spec] Unfudge .reduce test now that enough of .reduce is implemented to run it.
colomon \o/
19:41 cl58 left
colomon jnthn: on the whole, my problem right now isn't being blocked. it's having too many cool corners to explore. 19:42
:)
19:42 cl58 joined 19:43 hercynium joined, hercynium left, hercynium joined
colomon also only being able to type one-handed thanks to having a napping child in my lap. :) 19:43
jnthn: oh, there is still the ... thing 19:45
ng: (1, 3, 5 ... *).batch(10).perl.get
p6eval ng cb7f23: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class 'Parcel'␤current instr.: '!postcircumfix:<[ ]>' pc 314753 (src/gen/core.pir:41266)␤»
19:46 jonasbn joined, rindolf left 19:47 ewilhelm joined, iblechbot joined
ewilhelm pmichaud, did you post the code from your oscon lightning talk? 19:48
pmichaud ewilhelm: I don't remember if I did or not. But I certainly can :-) 19:49
ewilhelm that would be awesome. I seem to remember that it was quite an example of operators and other features.
19:49 araujo left
pmichaud www.pmichaud.com/2009/pres/oscon-perl6op 19:50
it's in there :-)
www.pmichaud.com/2009/pres/oscon-pe...lide5.html
19:50 araujo joined
ewilhelm thanks. Did you get my mail last month? 19:52
colomon oooo, that's nice (not in reference to pmichaud's classic example :)
pmichaud yes, but I've been very far behind on email 19:53
ewilhelm trying to put together a "what you've missed in the last several years" talk for a mostly ruby-using crowd
19:53 araujo left
ewilhelm if anybody has suggestions, that would be great. I thought I might work through the advent calendar as well. 19:54
19:54 dugg left, araujo joined 19:56 ShaneC2 joined, ShaneC1 left
pmichaud (advent calendar)++ 19:57
frettled (++)++ 19:58
pmichaud: so, ehrm, are you restful?
pmichaud getting there :-)
vacation was excellent, but lots of things stacked up while I was out :-| 19:59
frettled I know the feeling all too well.
I'm still trying to catch up with the consequences of Christmas.
TimToady phone 20:00
frettled hello?
20:00 cl58 left
frettled seems like a prank call ;) 20:02
pmichaud: I'm still enthused with ng 20:03
PerlJam wonders if there's a speech to text thingy that chromatic could use for the cabalcall
20:04 old_gray_bear left
PerlJam frettled: good. many people are enthused but slightly annoyed that ng isn't called master :) 20:04
("many" may have been too strong)
pmichaud ("slightly" may have been too weak :-)
frettled PerlJam: I have a very, very large ego, and I want it to be called master. My ego is so large it counts for ten people. 20:05
PerlJam that too :)
frettled: how big are your id and superego?
frettled PerlJam: Mu⁴
colomon ng: say 61731 * 61731 20:08
p6eval ng 9c17cb: OUTPUT«3810716361␤»
colomon ng: say
p6eval ng 9c17cb: OUTPUT«␤»
frettled ng: say 2²
p6eval ng 9c17cb: OUTPUT«Confused at line 1, near "say 2\x{b2}"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
frettled döh :)
I figured just now that writing the exponent like that would be really cool. 20:09
PerlJam idly wonders if there's a way to curry ** for all of those numeric superscripts.
(an "easy" way)
ng: say "\x{b2}" 20:10
p6eval ng 9c17cb: OUTPUT«Could not find non-existent sub &b2␤current instr.: '_block23' pc 110 (EVAL_1:50)␤»
PerlJam ng: say \x{b2}
p6eval ng 9c17cb: OUTPUT«Confused at line 1, near "say \\x{b2}"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
frettled ⁽¹²³⁴⁵⁶⁷⁸⁹⁾
colomon rakudo: say 61731 * 61731 20:11
p6eval rakudo 1d4928: OUTPUT«3810716361␤»
PerlJam colomon: were you hoping the answer to change? :)
colomon yes.
wish there was a sheepish smiley
frettled :-} 20:12
colomon trying to track down a Rat math problem, but it turns out to be a test glitch!
pugs_svn r29677 | colomon++ | [t/spec] Unfudge 22 tests thanks to jnthn++'s dispatch fix. Also fix a broken test. 20:19
20:19 payload joined
frettled PerlJam: I guess I'm just a bit tired, since I missed your comment regarding currying, and now that I saw it, I don't really understand what you're getting at :) 20:19
20:21 ruoso left
TimToady :】 20:22
dalek kudo/ng: eeff25d | (Solomon Foster)++ | src/core/Rat.pm:
Pull in the updated Rat operators from master.
colomon ng: say 1; 20:23
p6eval ng 9c17cb: OUTPUT«1␤»
frettled TimToady: there are two dots next to your phone handset? 20:24
PerlJam frettled: one way to get 2² to actually do the math is to define a postfix:<²> that takes an arg and squares it. This can be done with ** as $n ** 2. I was thinking of how one could automatically do that with all of the superscripts changing the 2 to the appropriate number.
colomon ng: (^1000).grep({my $b = $^a; ($b % 3 == 0) || ($b % 5) == 0}).reduce({$^a + $^b}).say 20:25
p6eval ng 9c17cb: ( no output ) 20:26
frettled PerlJam: yeees, but will it then work intuitively for expressions?
colomon > (^1000).grep({my $b = $^a; ($b % 3 == 0) || ($b % 5) == 0}).reduce({$^a + $^b}).say
233168
PerlJam frettled: it should 20:27
20:27 payload left, Chillance left
frettled So now we only need to support Knuth's ↑ notation ;) 20:28
That's a bit more tricky. 20:29
20:29 justatheory left
colomon ng: (^1000).grep({($^b % 3 == 0) || ($b % 5) == 0}).reduce({$^a + $^b}).say 20:29
p6eval ng 9c17cb: ( no output )
20:29 ashleyb joined
colomon dang it, still not fast enough 20:29
20:29 ash__ joined
frettled hmm 20:30
jnthn Dinner: nommed. 20:31
frettled PerlJam: Just in case you're not familiar with it: en.wikipedia.org/wiki/Knuth's_up-arrow_notation
jnthn Rakudo hacking!
colomon \o/
PerlJam ng: sub foo ($a,$b) { say "$a $b"; }; foo(3,5); my &x = &foo.assuming( b => 7);
p6eval ng 9c17cb: OUTPUT«3 5␤»
PerlJam ng: sub foo ($a,$b) { say "$a $b"; }; foo(3,5); my &x = &foo.assuming( b => 7); x(19);
p6eval ng 9c17cb: OUTPUT«3 5␤19 7␤»
frettled jnthn: so, do you like the idea of using ⁽¹²³⁴⁵⁶⁷⁸⁹⁰⁻⁺⁾ directly? 20:32
PerlJam ng: our &postfix:<²> := &infix:<**>.assuming(b => 2); say 2²; 20:33
p6eval ng 9c17cb: OUTPUT«Confused at line 1, near "our &postf"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
PerlJam ng: &postfix:<²> := &infix:<**>.assuming(b => 2); say 2²;
p6eval ng 9c17cb: OUTPUT«Confused at line 1, near "&postfix:<"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
PerlJam std: our &postfix:<²> := &infix:<**>.assuming(b => 2); say 2²; 20:34
p6eval std 29677: OUTPUT«ok 00:02 132m␤»
jnthn PerlJam: user defined ops don't work yet in ng. :-P
PerlJam i'm going to go with NYI for rakudo unless someone can show me different
ah, there we go 20:35
frettled heh
PerlJam jnthn: they work in master?
ash__ colomon++ jnthn++ so many commits in the last 24 hours, i take it ng is going well?
jnthn Yeah, should do.
PerlJam rakudo: our &postfix:<²> := &infix:<**>.assuming(b => 2); say 2²;
p6eval rakudo 1d4928: OUTPUT«Confused at line 10, near "\x{b2};"␤in Main (file <unknown>, line <unknown>)␤»
jnthn ash__: Not bad. :-)
colomon ash__: we are definitely making progress again. :) 20:36
mberends since the ng spectest count is rising strongly, I think it's time to give it its own docs/spectest-progress.csv and start doing those graphs again 20:38
phenny mberends: 09 Feb 22:47Z <jnthn> tell mberends sorry, connection dropped...I think use and require may need some review since imports need to be lexical etc - check with Pm.
colomon mberends++ # fine idea 20:39
20:39 ash___ joined
ash___ what % of tests does ng pass compared to master? 20:40
mberends it was about 18% recently
ash___ cool
20:41 hercynium left, ash__ left 20:42 justatheory joined
frettled PerlJam: (sorry, was AFK) - now, how about <⁽>, <⁾>, <⁺>, and <⁻>? :-> 20:42
20:42 cognominal left, cognominal joined
ash___ are nqp classes parrot classes once they get compiled? 20:46
jnthn ash___: Sorta, but no more than in the sense that Perl 6 ones are. 20:52
ash___: That is, we create a Parrot class behind the scenes, but it's dealt with by the metaclass.
ash___ yeah, but if you define a class in NQP for a parser, what kind of object is that? a parrot one? under the hood i mean 20:53
jnthn Under the hood, yes.
PerlJam at least until NQP gets some other backend :) 20:54
dalek kudo/ng: 54ca0fa | jonathan++ | t/spectest.data:
Add back three more test files that we've come to pass.
 20:56
colomon \o/ 20:57
jnthn Where I kinda want to end up is that we have "knowhow" as the primitive that the VM provides, and then ClassHOW is written in NQP or some such.
colomon: Small ones, but nice to have. :-)
ng: my $x; say $x.notdef
p6eval ng eeff25: OUTPUT«1␤»
pmichaud (spectest-progress.csv)
my intent is to (re)do list assignment and array fixes tonight, then switch ng branch to master shortly thereafter (probably tomorrow afternoonish) 20:58
if things go smoothly tonight I could do the branch rename tonight
ash___ whats the diff between notdef and defined?
20:58 macae left
colomon jnthn: I reckon every one that can added in is a Good Thing, because more tests running is more assurance we're not breaking stuff as we work. 20:58
jnthn ash___: !
pmichaud then I'll bring spectest up to date
jnthn colomon: Right. 20:59
mberends pmichaud++
pmichaud we'll see a big drop in spectest-progress.csv as a result... but that's reality :-)
mberends ng: my $t1=time;run "sleep 10";my $t2=time;say $t2-$t1 # sleep() without a function
p6eval ng eeff25: OUTPUT«10.0636529922485␤»
jnthn :-P
pmichaud ("time" is a function :-)
jnthn mberends: That's cheating! :-)
pmichaud (so is "run" :) 21:00
(so is "say" :)
jnthn pmichaud: Sounds like a good plan. ng is coming along quite quickly at the moment.
pmichaud jnthn: I agree, you all are doing great work. :)
jnthn pmichaud: Question.
mberends btw, we have no safe mode again
jnthn pmichaud: What's your plan on Match objects?
ash___ jnthn: i don't think i understand what you meant by !
pmichaud jnthn: (Match) I was planning to keep an approach similar to rakudo master... why? 21:01
jnthn ash___: Heh, that'll teach me to be lazy. ;-)
colomon ash___: he means notdef is !defined
pmichaud (keeping in mind that Match is different between nqp-rx and PGE)
jnthn ash___: It is just method notdef() { !self.defined }
:-)
colomon must admit he thought jnthn was excited at first...
jnthn pmichaud: I mean, at the moment postcircumfix:<[ ]> and <{ }> are not implemented by Regex::Match 21:02
ash___ ah, okay, i get it
pmichaud jnthn: they weren't implemented by PGE::Match either :-)
jnthn pmichaud: Are we going to subclass it and implement those in a subclass? Add a cheat that pokes them in?
pmichaud: Or treat them as foreign?
pmichaud jnthn: probably whatever was done in master applies :)
ash___ ng: "hi" ~~ /(.*)i/; say $/; say $0; # i can haz $0 works now?
p6eval ng eeff25: OUTPUT«Confused at line 1, near "say $0; # "␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤» 21:03
pmichaud we will subclass it though, yes.
jnthn ash___: You can't, I'm currently interrogating Pm on how to get us it back. ;-)
pmichaud looks like in master we treated them as foreign aggregates
jnthn OK. 21:04
pmichaud I'd say do the same here for now
jnthn That can work too.
Might try and pop that in tonight.
It's be Nice To Have.
pmichaud ng: "hi" ~~ /(.*)i/; say $/[0];
p6eval ng eeff25: OUTPUT«Can't postcircumfix:<[ ]> foreign objects yet.␤current instr.: '!postcircumfix:<[ ]>' pc 316355 (src/gen/core.pir:41714)␤»
pmichaud hmmmm
well, I expect to also deal with grammars in the next 24h 21:05
ash___ $/<something> isn't working yet either
pmichaud depends on what else comes up around here, though
jnthn Sure.
There's loads of little bits to work on. :-)
pugs_svn r29678 | jnthn++ | [t/spec] Couple of unfudges in a test file for tests that we now pass that master didn't. 21:06
colomon we have a week until the next release?
jnthn Is that all?!
pmichaud week + day :)
jnthn Ooh, yes.
pmichaud I know I haven't been very P6-productive lately; apologies to the group for that. :-| 21:07
TimToady maybe postfixes should have a ! metaoperator too
$x.!defined
$x!.defined
pmichaud !$x.defined :-) 21:08
ash___ (insipired by ruby) $x.defined?
TimToady !really_long_expression.defined.oops_there_was_more
mberends pmichaud: (P6-productive) that's cost you 0.1% of your forgiveness capital ;)
21:09 synth left, synth joined
mberends spectest-progress.csv update: 115 files, 3235 (20.4% of 15828) pass, 0 fail 21:10
jnthn eek
.sub '!FAIL' .param pmc args :slurpy $P0 = new ['Undef'] .return ($P0)
.end
Undef is no more. :-) 21:11
jnthn changes it cheat differently.
TimToady it's undefined...
jnthn TimToady: Right
PerlJam unMu
jnthn TimToady: But there ain't an Undef type.
21:11 cognominal left 21:12 cognominal joined
[particle] Undef is undefined 21:14
it's pushing up NaNs
ash___ undef !== Undef basically? 21:15
TimToady std: undef 21:17
p6eval std 29678: OUTPUT«===SORRY!===␤Unsupported use of undef as a value; in Perl 6 please use something more specific:␤ Mu (the "most undefined" type object),␤ an undefined type object such as Int,␤ Nil as an empty list,␤ *.notdef as a matcher or method,␤ Any:U as a type
..constraint␤ or …
21:18 takadonet left
TimToady phenny: tell masak Positive lookaheads must be compiled at compile time, or they cannot participate in LTM as specced, and as STD implements. 21:27
phenny TimToady: I'll pass that on when masak is around.
jnthn rakudo: say $*PID 21:30
p6eval rakudo 1d4928: OUTPUT«Contextual $*PID not found␤␤»
jnthn ng: say $*PID
p6eval ng 54ca0f: OUTPUT«␤»
21:33 xomas_ left 21:34 payload joined
dalek kudo/ng: df1e34e | (Martin Berends)++ | src/core/system.pm:
[core/system.pm] add a sleep() function that does and returns fractional seconds
 21:37
21:38 uniejo left
jnthn cognominal: about? 21:41
cognominal: oops, sorry
colomon: about?
colomon yes
jnthn I just tweaked FAIL to create a Failure with an unthrown exception. 21:42
colomon still rocking my son. :)
jnthn :-)
It breaks a couple of tests, though I think legitimately...
One is range.t though
"Use of type object as a value"
In test 78
colomon which range.t? 21:43
jnthn The others are all in IO, where it tries to use $*PID
S03-operators/range.t
AFAICT, looking up a contextual var that isn't defined and testing it for definedness is OK.
But trying to use one that doesn't exist is meant to be an error. 21:44
As in, it throws.
Guess we were getting away with that before now though.
colomon .first works on ranges is the test?
pmichaud contextual vars that haven't been set return a Failure, iirc 21:45
normally the Failure should be "Use of undefined contextual $*FOO..."
jonasbn pmichaud: ping
pmichaud jonasbn: pong
jnthn pmichaud: Right, if you then try to use it though, that failure throws, yes?
pmichaud: So
if $*FOO.defined { ... } # totally fine, no exception
say $*FOO; # oh noes 21:46
colomon is very confused
jonasbn pmichaud: you will be in Copenhagen from Saturday to Wednesday?
jnthn colomon: I think the test that I've made explode was todo's actually.
colomon: Yes, it is the .first one 21:47
colomon I don't see any contextual variables?
jnthn #?rakudo todo '.first suffers from the bug where code blocks get the wrong $_'
colomon: Ah, sorry for the confusion.
colomon: It gives a failure object here due to use of proto-object as a value.
colomon: Which now throws.
colomon oh!
jnthn Sorry!
colomon I see
jnthn :-)
colomon either skip it or change $_ to $^a 21:48
that's a good change, really (the fail, I mean) it makes these problems stand out 21:49
jnthn colomon: I skipped it, and the rest of the test file is fine.
colomon: $*PID I can't even implement as it seems Parrot doesn't provide a way to give us it! 21:50
pmichaud jonasbn: I'm arriving Copenhagen on Thursday, flight departs Copenhagen early Wednesday
jonasbn: it's not too late for me to change that schedule, if desired :)
jonasbn okay
no no please don't
I just need to book hotel
pmichaud I decided that my trying to make it to Dutch Perl Workshop was going to make the trip too long for me
colomon jnthn: how was master doing PID? 21:51
jonasbn out of the 5 people coming, you are the only one coming Thursday, so you might have to spend time with me :)
pmichaud jnthn: ($*FOO failure) yes
jonasbn pmichaud: how old are your kid(s)?
pmichaud jonasbn: well, normally the first day I like to catch up on sleep :)
21:51 SmokeMachine left
pmichaud but yes, I'll have some time on Thur and/or Fry 21:52
*Fri
kids are 12 and 8
jonasbn okay, mine are almost 3 and just 5, so you can cope
jnthn colomon: $*PID just gave undef in master and we got away wiht it!
colomon: Or something.
:-)
pmichaud anyway, I figured that if I arrived on Thur, I'd be around for the first day of the OSD conference, in case we want to have any presentations then 21:53
jnthn colomon: I'm going to Cheat Horribly and just set it to 0.
colomon jnthn++
21:53 eternaleye joined
jonasbn I will be at the conference too, I am not sure about the Perl track I think we will save it for Saturday, but lets see, now we have the option at least 21:54
pmichaud jonasbn: okay.
jonasbn: I'm sure I'd also benefit from just being able to hack on my own for a day as well :) 21:55
jonasbn: or get in some more sightseeing of copenhagen :)
jnthn Copenhagen is a nice city to sightsee in. :-)
jonasbn pmichaud: we can arrange that too, either hotel or I can set you up at my $job
it is really cold here now 21:56
jnthn jonasbn: How cold is "really cold"?
pmichaud hotel or $job works for me :) 21:57
jonasbn jnthn: -3 degrees celcius
jnthn colomon: That seems to get things passing again. :-)
jonasbn: Ah, same-ish as here. :-)
jonasbn well for walking around all day, it is a bit cold, especially if you come in your hawaii shirt, sandals and bermuda shorts 21:58
jnthn jonasbn: I took one poor visitor sightseeing in Trencin last month, when it was -10C!
colomon: Gah, spoke too soon...
colomon ditto (25F), and this isn't really cold by our standards...
pmichaud -3 celcius isn't that cold :)
it was -3 celsius here this morning :)
jnthn colomon: oh gah, another one uses $*OS! 21:59
pmichaud hmmmm, if I left on Tuesday, I could flight to AMS and perhaps make it to Dutch Perl Workshop
(cheaper flights, too :)
colomon jnthn++ # suffering the pain of doing the right thing
dalek kudo/ng: c36e0f4 | (Martin Berends)++ | t/spectest.data:
[t/spectest.data] add S29-context/sleep.t (passes all 4 tests)
 22:00
jonasbn pmichaud: all of the others are not coming until Saturday 22:01
Kvorg silly question ... what is the accepted way to iterate the stdio, i.e. while (<>) from perl5?
jnthn colomon: Gonna put $*OS and $*OSVER back "properly"
pmichaud jonasbn: right. So, would it be good to make sure I'm there as of Saturday morning (in case others are late or late-ish), or should I try to arrive the same time as others?
mberends ng: say sleep(0.7) 22:02
p6eval ng df1e34: OUTPUT«0.724602937698364␤»
jnthn pmichaud: I suspect mberends++ and I will be there very early. :-)
mberends: yay, sleep!
pmichaud jnthn: okay, that would be good then. I don't know how awake we'll all need to be for conference/hackathon on saturday, though :)
jonasbn pmichaud: what ever is easiest for you, if you want to come here Thursday, it is fine with me, I just need to book hotels
jnthn pmichaud: I won't be hugely awake that day, I suspect. 22:03
pmichaud: Well, it all hangs off how good the coffee is I guess. :-)
pmichaud jonasbn: well, I plan to book the ticket in the next 8 hrs, so I can definitively let you know then
(my reservation expires 7 hours and 55 minutes from now :) 22:04
jonasbn okay, then I will book hotels tomorrow
pmichaud jnthn/mberends: any opinion on whether I should try to make it to nlpw?
(and I need to check with my wife also) 22:05
mberends pmichaud: we can arrive earlier to make a mini-Hackathon is necessary. The schedule for the entire week is open :)
*if
pmichaud how/when to get from AMS to nlpw? 22:06
i.e., if I flew to AMS, how would I make it to nlpw?
mberends we'll have a big spacious enough hired car
jnthn Yeah
mberends so we pick you up
pmichaud what day should I arrive AMS? Wed or Thu?
(iirc, nlpw is Fri) 22:07
mberends (gotta crush jnthn's @stuff a little, but np ;)
jnthn Yes, it's Fri
PerlJam Kvorg: for lines() { ... }
jnthn oh noes my stuffs!
pmichaud oh, I don't have a lot of stuff. Everything I'll be bringing fits under the airline seat :)
Kvorg PerlJam: Thanx!
PerlJam Kvorg: though, I don't think that actually works at the moment
jnthn pmichaud: The only thing to consider is that we'll be doing an overnighter from nlpw to Copenhagen. 22:08
pmichaud: And I dunno how that plays out with jetlagged you. :-)
pmichaud jnthn: that's not too big a concern for me I don't think.
jnthn OK. :-) 22:09
pmichaud (as long as I'm not driving, that is :)
Kvorg hehe, no it does not: Symbol '$line' not predeclared in <anonymous> (test.pl:2) in Main (file <unknown>, line <unknown>)
jnthn It woudln't be for me either, but I know some people who get pretty beaten up with jetlag. :-)
pmichaud yes, that's why I normally like to arrive a day early to catch up on sleep :-)
Kvorg i always thought that should be the first think i test when i build a new rakudo, but either it never works, or i never know the correct syntax of the moment.
kinda the most basic think to expect from a perl at any day ...
jnthn $*IN.lines does (should) work 22:10
mberends the NLPW time itself will probably be one track of talks each under 20 mins, similar to www.perlworkshop.nl/nlpw2009/talks 22:11
PerlJam Kvorg: note that I said lines(), not $line
pmichaud mberends: are you and jnthn already traveling from ams to nlpw? or is that a special trip? or otherwise "on the way"?
mberends it's about a 1 hour deviation, very easy to splice in 22:12
pmichaud when should I be in AMS for the splice?
jnthn In Texas terms, it's just around the corner.
;-)
pmichaud ("one hour" == "across town" in Texas terms :) 22:13
ash___ ng: say $*IN.lines.defined;
p6eval ng df1e34: OUTPUT«1␤»
22:13 barney left
pmichaud or, put another way, when were you and jnthn planning to head to nlpw? 22:13
jnthn ng: for $*IN.lines { .say }
ash___ can you set $*IN to something else? like a file?
p6eval ng df1e34: OUTPUT«Land der Berge, Land am Strome,␤Land der Äcker, Land der Dome,␤Land der Hämmer, zukunftsreich!␤Heimat bist du großer Söhne,␤Volk, begnadet für das Schöne,␤vielgerühmtes Österreich,␤vielgerühmtes Österreich!␤␤Heiß umfehdet, wild umstritten␤liegst dem Erdteil du inmitten,␤einem
..stark…
jnthn ash___: sure
mberends we had planned to travel Wednesday night to arrive in NL Thursday morning, but it's flexible (speaking on behalf of jnthn++)
ash___ cool, just didn't know if that functionality worked yet 22:14
jnthn mberends, pmichaud: I'm flexible.
pmichaud that might work
jnthn ash___: We're not running the test yet I don't think...
PerlJam ash___: $*IN is just a scalar that happens to already have a nice value. You're free to change that value
jnthn Well, that test file output is hilarious 22:15
Kvorg my bad, actually it does work, but it seems to wati for the end of file: for $*IN.lines { say "Read the line: $_"; }
pmichaud looks like I could arrive AMS at 15h55
Kvorg (should I go for nopaste for little bits like that?)
jnthn Kvorg: In master, yes?
PerlJam Kvorg: aye, that's a laziness issue that's being worked upon.
pmichaud (Wednesday, 15h55) would that be early enough?
jnthn Oh wait, for isn't lazy yet in ng either...
pmichaud: That could possibly be workable if mberends++ is ammeanable to us setting off at $EARLY-HOUR on Wednesday from here. 22:16
Kvorg and so does for $*IN.lines { .say }, with the same caveat
mberends pmichaud, jnthn: cool, we can do that
pmichaud jnthn: I'd be okay with hanging around AMS for a bit as well, if you're arriving later
also it'd be good to not plan that my flight will be on-time :) 22:17
Kvorg what would one have to do to get non-buffered read from a terminal? is it possible with such a construct? (perhaps a pointless question) 22:18
pmichaud so, if you two are arriving later in ams, that'd be no big deal. I might even temporarily rent a place to sleep until you arrive :)
22:18 Su-Shee left
mberends pmichaud: on Wednesday there's no pressure on anyone to be on time :) 22:18
AMS has free WiFi in some corners 22:19
pmichaud okay. I'll mull it over, discuss it with wife, and book flights tonight and let everyone know what's up tomorrow :)
jnthn OK. :-)
colomon: Pushed the bunch. 22:20
colomon jnthn: commit test changes as well, or did I miss that earlier?
jnthn d'oh!! 22:22
colomon: done, thanks.
dalek kudo/ng: 3a94b96 | jonathan++ | src/cheats/setup-io.pm:
Implement $*OS, $*OSVER and a hugely cheating and lying $*PID.
kudo/ng: fe967ce | jonathan++ | t/spectest.data:
We now pass S02-names_and_variables/contextual.t again.
kudo/ng: 446afaf | jonathan++ | src/cheats/fail.pir:
Cheat less with failures. They actually throw now.
pugs_svn r29679 | jnthn++ | [t/spec] Turn a todo into a skip - we fail harder now.
22:25 Su-Shee joined
colomon ooo, so close to running 3600 tests on my system... 22:27
on the other hand, abs.t fails now. 22:28
22:29 jonasbn left
jnthn colomon: Oh? 22:29
colomon: I get 8 passing TODOs here. :-/
mberends be there dragons if I attempt to bring back %*ENV ? 22:30
pugs_svn r29680 | colomon++ | [t/spec] Change to-do to skip because of harder fails.
jnthn mberends: RAAR
mberends: Hash changed some
mberends: But probably not too many dragos.
colomon jnthn: huh, but it was definitely failing here. 22:31
I'm going to look at the to-dos.
jnthn colomon: If it's numbery stuff, it could be platform specific annoyances.
colomon: fwiw, I sign.t tends to fail here too. 22:32
colomon hmmm...
Su-Shee moritz_: the bertrand meyer book is extremely cool.
colomon ng: say 0.WHAT; say 0.abs.WHAT 22:33
p6eval ng c36e0f: OUTPUT«Int()␤Num()␤»
jnthn Gah. For the last two hours I've done this: loop { boil_kettle(); hack(); realize_its_no_longer_warm(); } 22:34
Tene hehehe 22:36
pugs_svn r29681 | jnthn++ | [t/spec] Remove bogus returns that's nothing to do with the test anyway. 22:37
jnthn yay
$tea.slurp()
mberends jnthn: snap! 22:38
PerlJam ng: "Nice Day Isn't It?".lc.comb.sort.join.say 22:39
p6eval ng c36e0f: OUTPUT«Method 'sort' not found for invocant of class 'GatherIterator'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
PerlJam Hrm.
jnthn I wishes we had sort.
We're missing some tests for not doing so.
PerlJam rakudo: "Nice Day Isn't It?".lc.comb.sort.join.say 22:40
p6eval rakudo 1d4928: OUTPUT« '?acdeiiinnstty␤»
22:41 cotto_working left
PerlJam rakudo: "Nice Day Isn't It?".lc.comb.uniq.sort.join.say 22:41
p6eval rakudo 1d4928: OUTPUT« '?acdeinsty␤»
jnthn Oh.
:-/
PerlJam I don't understand the error message "method not found on GatherIterator". what's up with that? 22:42
pmichaud looks to me like .comb is returning a GatherIterator 22:44
might need to be a List with a GatherIterator
(it's a question of whether you want to treat an Iterator as a scalar object or as something that acts like a list)
22:45 Su-Shee left
PerlJam ng: "foo".comb.sort 22:45
p6eval ng c36e0f: OUTPUT«Method 'sort' not found for invocant of class 'GatherIterator'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
PerlJam ng: "foo".comb.uniq
p6eval ng c36e0f: OUTPUT«Method 'uniq' not found for invocant of class 'GatherIterator'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
PerlJam ng: "foo".comb.join
p6eval ng c36e0f: ( no output )
22:46 nacho left
PerlJam 1 out of 3 ain't bad ;) 22:46
pmichaud I suspect .join is doing a .list on its invocant first
anyway, most of that still needs cleaning up, I suspect.
jnthn ng: my $aref = [0]; $aref[0]++; 22:48
p6eval ng c36e0f: OUTPUT«sh: ./perl6: No such file or directory␤»
22:48 k23z__ left
jnthn pmichaud: Above gives Cannot assign to readonly value 22:48
pmichaud: Is that something that'll get cleared up in the list/array re-doing you've got planned? 22:49
s/list/list assignment/
22:49 ash__ joined
dalek kudo/ng: 2561333 | jonathan++ | src/core/EnumMap.pm:
.Num and .Int on EnumMap, and thus Hash.
 22:51
22:51 ash___ left 22:54 kst joined
pmichaud jnthn: yes. 22:57
colomon oooo, Enums! 22:59
afk # orchestra
jnthn pmichaud: yay. 23:00
23:01 dual left
jnthn pmichaud: Hashes have the same issue. 23:04
23:04 hercynium joined
jnthn If the fix that does it for arrays doesn't help there, I can at least spot it and apply the same trick for the hashes though :-) 23:04
23:05 cotto_working joined
ash__ how do you make a capture? isn't it \? like my $a = \$foo; ? 23:07
23:07 nihiliad left
jnthn \(...) should do it 23:14
dalek kudo/ng: dacbea0 | jonathan++ | (2 files):
Add back Signature.perl, with a couple of minor tweaks for it to work under ng.
ash__ ng: my $a = 5; my $b = \($a); 23:15
p6eval ng fe967c: OUTPUT«Confused at line 1, near "my $b = \\("␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
ash__ does := not work in ng? its also giving me an error
ng: my $a := 5; 23:16
p6eval ng fe967c: OUTPUT«Confused at line 1, near "my $a := 5"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (src/stage0/HLL-s0.pir:328)␤»
23:17 ashleyb left
jnthn ash__: Not yet. 23:18
23:18 ashleyb joined 23:19 cognominal left
pmichaud := nyi in ng 23:24
jnthn := is only sorta i in master. :-)
pugs_svn r29682 | jnthn++ | [t/spec] Skip two tests due to [0] constructing a read-only array in ng. 23:25
23:27 iblechbot left
pugs_svn r29683 | jnthn++ | [t/spec] Turn a couple of todos into skips since ng hardfails 'em; they're all RT tracked, anyways. 23:31
23:31 pmurias left
dalek kudo/ng: 8a4c410 | jonathan++ | src/Perl6/Actions.pm:
Fix an ickle bug in named parameter handling.
 23:32
kudo/ng: cf9f32f | jonathan++ | src/builtins/Proxy.pir:
my %oh; %oh<lol>.notdef should be true; make it so.
kudo/ng: 5b5e6a6 | jonathan++ | t/spectest.data:
S06-signature/named-parameters.t passes again.
jnthn std: whatever( 'a' p5=> 'b', 'c' p5=> 'd' ); 23:34
p6eval std 29681: OUTPUT«===SORRY!===␤Unable to parse argument list; couldn't find final ')' at /tmp/6IOXfE1t73 line 1:␤------> whatever( 'a' ⏏p5=> 'b', 'c' p5=> 'd' );␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤FAILED 00:01 108m␤»
23:36 kidd` joined, kidd` is now known as rgrau
pugs_svn r29684 | jnthn++ | [t/spec] p5=> is dead. 23:36
23:43 ash__ left 23:47 drbean joined
dalek kudo/ng: 2d8d82a | jonathan++ | t/spectest.data:
One more passing test.
 23:49
23:56 Limbic_Region 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!