»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
Radvendii no. i'll look it up 00:00
benabik perlgeek.de/blog-en/perl-6/dissecti....writeback
Radvendii HAHAHA
win
dalek ecza: 3e212e9 | sorear++ | lib/C (2 files):
Implement a crude approximation of immutable objects, use it in does
shinobicl___ perl6: class C { has Int $.v = 0; submethod BUILD() {say "Constructing with " ~ $.v}; }; my C $A=C.new(); 00:01
p6eval rakudo 2b5363: OUTPUT«use of uninitialized value of type Int in string context␤Constructing with ␤»
..niecza v13-327-g5a9c36f: OUTPUT«===SORRY!===␤␤Virtual call $.v may not be used on partially constructed object at /tmp/F4LJYPYujB line 1:␤------> hod BUILD() {say "Constructing with " ~ ⏏$.v}; }; my C $A=C.new();␤␤Potential difficulties:␤ $A is decl…
..pugs b927740: OUTPUT«Constructing with 0␤»
00:26 Radvendii left 00:36 Chillance left 00:38 replore joined
[Coke] sorear: if we took rakudo's temporal.pm files, how would you want to include them into niecza? doesn't look like we have any non-CORE perl6 files. (but we do have some non-CORE .cs files) 00:50
00:58 LlamaRider joined 00:59 molaf_ joined
sorear [Coke]: are they import modules or part of the setting? 00:59
[Coke] it looks like they are separate files that are included into the setting.
sorear then they'd have to be incorporated into CORE.setting 01:00
[Coke] inline, or is there a way to just refer to the external files? 01:01
sorear inline
[Coke] OK.
01:02 molaf left 01:03 cognominal_ joined 01:06 cognominal left 01:16 shinobicl___ left 01:19 Radvendii joined
Radvendii is there a shorthand for "@a[@b[0]][@b[1]]"? 01:20
benabik nom: my @b = 7,8,9; say @b[0,1] 01:21
p6eval nom 2b5363: OUTPUT«7 8␤»
Radvendii A.K.A using vectors
benabik: I'm talking about a multidimensional array
benabik Radvendii: Yeah, I realized I didn't know how to split that to use in a multi-dim... 01:22
Radvendii oh
okay
dalek ecza: 0da7a8d | sorear++ | / (6 files):
Allow "constant" declarations to be folded as constants
 01:23
sorear Radvendii: what is @b? 01:24
Radvendii nom: my @a=[0,1],[2,3]; my @b=0,1; say @a[@b[0]][@b[1]]
p6eval nom 2b5363: OUTPUT«1␤»
Radvendii something like that
except that i'm doing it over and over again
so it would be more useful to be able to say "say @a[[@b]]" or something... 01:25
sorear You haven't explained what @b is
Radvendii what do you mean... it's an array holding the location of the information in @a 01:26
01:26 swarley joined
sorear can you guarantee that @b will always have exactly as many elements as dimensions in @a? 01:26
the code you presented earlier doesn't seem to require that 01:27
[Coke] nieczaa: my int $a = 32;
Radvendii you can assume that if that makes the solution simpler
sorear generally, if you want help with a problem, you should present it in English
it's a LOT easier that way 01:28
[Coke] niecza: my int $a = 32;
p6eval niecza v13-329-g3e212e9: OUTPUT«===SORRY!===␤␤Malformed my at /tmp/gqHrzQ1HFT line 1:␤------> my ⏏int $a = 32;␤␤Parse failed␤␤»
Radvendii sorry, i was trying, but i didn't know how to render it exactly.
sorear anyways, if @a was a real multidimensional array (NYI), you could use @a[||@b]
LlamaRider S09 says something about ||
aaaaaand I'm late :D
Radvendii what do you mean "real"?
sorear my @a[3;4] 01:29
LlamaRider sorear: rakudo doesn't seem to support || at the moment though
sorear LlamaRider: that's what NYI means, yes 01:30
[Coke] nothing supports multidim arrays atm.
NYI == not yet implemented.
LlamaRider ah, those acronyms... sorry
[Coke] sorear, you can't say "use english" and then not define terms. ;)
Radvendii oh! we can do @a=(0,1),(2,3) now?
[Coke] niecza: say sprintf "%20d", 30.2 01:31
p6eval niecza v13-329-g3e212e9: OUTPUT« 30␤»
Radvendii without using references?
and by "now" i mean when they implement it?
[Coke] sorear: copying Temporal into the setting, this sub dies at the !! 'Z'; -
gist.github.com/1659855
sorear No, you can't use that syntax to initialize multi-dim arrays 01:32
sorear seems to be in a bad mood now, sorry
Radvendii alright, i'll look it up when they actually implement it...
sorear that is some strange code formatting, [Coke] 01:33
[Coke] # 01/22/2012 - niecza at 99.72%
(we're slipping).
sorear I wonder if putting parens arround the sprintf args would help it parse?
[Coke] also, we have more failures than yesterday.
sorear [Coke]: why?
Radvendii are we going to be able to make our own sigils at some point? (for important enough classes) 01:34
[Coke] feather.perl6.nl/~coke/niecza_summary.out
sorear the answer to all questions like that is "yes"
[Coke] failures are at the bottom; and that's missing the last commit or 2. 01:35
Radvendii cool. :P no rush, and i really do appreciate all the effort you guys are putting into this language.
[Coke] what is ObjAt? (found in nom source) 01:37
ah, perlcabal.org/syn/S02.html#The_ObjAt_Type 01:38
sorear ahahaha I know why env.t passes for me and fails for you
[Coke] sorear: can't use rakudo's Temporal.pm without ObjAt.
sorear: yeeeees?
01:38 Radvendii left
LlamaRider Radvendii: a (perverted?) way of getting what you want right now would be: 01:38
rakudo: my @a=[0,1],[2,3]; my @b=0,1; my @c = @a; @c = @c[$_][] for @b; say @c;
p6eval rakudo 2b5363: OUTPUT«1␤»
sorear [Coke]: you have TERMCAP set in your environment 01:39
[Coke] Aye.
sorear env.t relies on .perl/eval to move data
niecza's Str.perl doesn't escape all the funny characters in your $TERMCAP properly
and the eval dies
LlamaRider he quit just as I wrote oO unlucky 01:40
sorear "I only have reasonable characters in my %ENV"
[Coke] sorear: ()'s helped with the !! ?? parsing. 01:41
01:41 Radvendii joined
Radvendii hey, "method BUILD" is used to initialize variables right? 01:42
sorear no
first, it has to be _sub_method BUILD
second, BUILD is used *only* for classes and instance setup 01:43
Radvendii but this is for a class... what else would one think to use it for? 01:44
sorear anything else you want to do at construct time
dalek ast: e6549e8 | sorear++ | S14-roles/basic.t:
[S14-roles/basic] Do not try to mutate 3
sorear [Coke]: this gets us 35 tests back, ~0.18% 01:45
Radvendii i dont understand... so i would or wouldnt use it to initialize an instance variable?
sorear you should ideally use initializers to initialize instance variables
01:46 LlamaRider left
Radvendii is that like saying "has $a=5"? 01:47
sorear yes
Radvendii ah 01:48
okay
01:51 Trashlord left 01:52 Radvendii left, Trashlord joined 01:54 shinobicl___ joined 01:59 whiteknight left 02:03 thou joined
shinobicl___ rakudo: class C { has $!val; method new(Int $x){ say $x } }; my $var = C.new(100); 02:11
p6eval rakudo 2b5363: OUTPUT«100␤»
shinobicl___ rakudo: class C { has $!val; method new(Int $x){ say $x } }; my C $var = C.new(100);
p6eval rakudo 2b5363: OUTPUT«100␤Type check failed in assignment to '$var'; expected 'C' but got 'Bool'␤ in block <anon> at /tmp/KGDJ1lBWbJ:1␤ in <anon> at /tmp/KGDJ1lBWbJ:1␤»
shinobicl___ is that a bug? 02:12
02:21 shinobicl___ left 02:33 swarley left 02:55 drbean left 02:56 drbean joined 03:12 drbean left 03:13 drbean joined 03:26 orafu left 03:27 orafu joined 03:42 Patterner left 03:44 Psyche^ joined, Psyche^ is now known as Patterner 03:55 drbean left 03:56 drbean joined 04:35 thou left 04:48 birdwindupbird joined 04:50 araujo left 05:04 jjore joined 05:10 Trashlord left 05:12 xinming left 05:31 thou joined
moritz good morning 05:33
06:07 araujo joined, araujo left, araujo joined 06:21 koban joined, koban left
moritz rakudo release uploaded and tagged 06:38
masak morning, #perl6 06:39
moritz++
shinobicl___: no, it's not a bug. 'say $x' returns a Bool, which you then try to assign to $var typed as C. boom. 06:40
masak --> $dayjob
06:41 sftp_ left 06:51 wtw_ joined 06:58 sftp_ joined 07:15 overrosy left 07:22 overrosy joined 08:01 replore left 08:02 kaleem joined 08:13 GlitchMr joined
sorear o/ GlitchMr 08:13
moritz++
dalek kudo/nom: 63663bd | moritz++ | docs/release_guide.pod:
correct release date and copyright in release_guide.pod
 08:30
08:35 drbean left
tadzik 'morning 08:36
GlitchMr Hi (late) 08:37
moritz \o tadzik
08:37 drbean joined
moritz just noticed that he did something stupid 08:37
the released rakudo wants parrot 3.11
since the PARROT_REVISION in nqp had a full release number in it, I thought it had already been bumped 08:38
and I didn't notice that the newest parrot is 4.0, not 3.11
08:41 mj41 joined
dalek ar: cc390ba | moritz++ | skel/docs/announce/2012.01:
small fixes to the release annoucement

turns out we accidentally ship parrot 3.11, not 4.0, unless somebody fixes it		 08:45
ecza: 155a5c1 | sorear++ | / (5 files):
Avoid excess subcalls when a default or type constraint (sometimes) is a compile-time constant
 08:49
sorear (yes, I know Rakudo got to that optimization first) 08:50
moritz you started later :-)
moritz had a weird idea the other day... 08:55
somebody could write a script that searches for tests that are todo'ed or skip'ped for both rakudo and niecza
and compile a list of them
because they are much more likely to be dubious than the average test
then we could review those tests more efficiently
sorear I recommend basing it off test_summary, and finding 'matching tests' by file/index pairs in the TAP 08:56
nom: sub foo($) { }; say &foo.signature.params.[0].name 08:57
p6eval nom 2b5363: OUTPUT«␤»
sorear nom: sub foo($) { }; say &foo.signature.params.[0].name.WHAT
p6eval nom 2b5363: OUTPUT«Str()␤»
sorear nom: sub foo($) { }; say &foo.signature.params.[0].name.perl
p6eval nom 2b5363: OUTPUT«""␤»
sorear nom: sub foo($bar) { }; say &foo.signature.params.[0].name.perl
p6eval nom 2b5363: OUTPUT«"\$bar"␤»
dalek ecza: 677cf15 | sorear++ | src/ (2 files):
mergeback
sorear tomorrow's agenda: major additions to Sub introspection API. Time permitting, also add MAIN 09:03
(also, first day of spring classes)
sorear -> sleep
jnthn o/ folks 09:16
jnthn got grabbed the moment he walked through the door at $dayjob today :)
moritz: (3.11) main worry is that we may miss cygwin build fixes put in to Parrot 09:17
moritz jnthn: oh. Should I make a 2012.01.1 point release that depends on 4.0? 09:18
jnthn moritz: Maybe hold on a bit and see if not_gerd appears and ask. 09:22
Or maybe somebody on #parrot knows if they got cygwin fixes.
moritz jnthn: just read the logs, the commits went in after the release 09:27
jnthn moritz: After 4.0? 09:28
moritz jnthn: yes
jnthn OK
moritz: I just looked over the commits between 3.11 and 4.0. Nothing jumped out at me as making it worth re-cutting the release. 09:32
09:32 thou left 09:56 dakkar joined 10:12 domidumont left 10:13 domidumont joined 10:29 daxim joined 10:36 fhelmberger left, fhelmberger joined
moritz jnthn: at some point I'd like to discuss with you how to discuss typed exceptions from the MOP 10:43
s:2nd/discuss/throw/
10:48 dbr joined
moritz hm, maybe $*W.throw simply works? 10:49
10:54 lutok left
jnthn moritz: Can't really rely on a $*W being available. 10:58
moritz: What if somebody is using the MOP at runtime, etc.
moritz jnthn: right. Urks. 10:59
hm 11:00
jnthn Well, it'll need a circularity saw somewhere, because we're defining the MOP that says how the Exception classes work. :)
moritz so we need something that at compile time uses $*W, and at run time searches the caller's lexpad, and if both fails, dies the old way without typed exceptions 11:01
does that sound about right?
jnthn It gets "worse" in so far as we probably need to throw typed exceptions from ops in some cases too.
moritz you mean C-level ops? 11:02
jnthn Yeah...I think some of those throw.
Or think about
nom: class A { has $.a }; A.a
moritz indeed that's worse, but thankfully those aren't covered in my grant :-)
p6eval nom 63663b: OUTPUT«Cannot look up attributes in a type object␤ in method a at src/gen/CORE.setting:1384␤ in block <anon> at /tmp/0yJ3JndYqf:1␤ in <anon> at /tmp/0yJ3JndYqf:1␤»
jnthn moritz: That may be, but I'd rather think up a mechanism that'll let us get those at some point too :) 11:03
moritz jnthn: sure, if it turns out that we can cover both with the same approach
jnthn moritz: Lemme think it over some.
moritz jnthn: ok, no hurry. Still plenty of LHFs in the actions and maybe setting 11:04
nom: class A { $^x } 11:11
p6eval nom 63663b: ( no output )
jnthn Hm. :) 11:12
moritz nom: do { $^x }
p6eval nom 63663b: OUTPUT«===SORRY!===␤Cannot use placeholder parameters in this kind of block at line 1, near ""␤»
moritz I guess the first one should die with the same error, no?
jnthn I guess
Though
nom: role R { say $^x }; 'lol' but R[42]
p6eval nom 63663b: OUTPUT«None of the parametric role variants for 'R' matched the arguments supplied.␤No applicable candidates found to dispatch to for '_block1400'. Available candidates are:␤:(Mu )␤␤ in specialize at src/gen/Metamodel.pm:1810␤ in specialize at src/gen/Metamodel.pm:1480␤ i…
jnthn could argue that one should be allowed ;) 11:13
Looks like package_def just doesn't go looking at if any placeholder signature was set on the block, though.
Should be an easy fix.
moritz tries to fix it 11:14
looks easy indeed 11:15
11:26 Exodist left 11:31 Exodist joined 11:58 Woodi left, Woodi joined 11:59 Vlavv` left 12:00 Vlavv` joined
moritz niecza: class A { $^x } 12:02
p6eval niecza v13-332-g677cf15: OUTPUT«===SORRY!===␤␤Placeholder variable $^x may not be used here because the surrounding package block takes no signature at /tmp/LwCSTfwG4B line 1:␤------> class A { ⏏$^x }␤␤Unhandled exception: Check failed␤␤ at /home/p6…
dalek kudo/nom: 3ddd2fb | moritz++ | src/ (2 files):
throw typed exception on illegal use of placeholder parameters

also catches class A { $^x }, which used to live errornously
moritz niecza: do { $^x }
p6eval niecza v13-332-g677cf15: OUTPUT«===SORRY!===␤␤Placeholder variable $^x may not be used here because the surrounding block takes no signature at /tmp/EcwqtNwCuw line 1:␤------> do { ⏏$^x }␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/b…
moritz niecza++ 12:03
dalek ast: fbb78ed | moritz++ | S32-exceptions/misc.t:
test for X::Placeholder::Block
 12:04
moritz niecza: role a { $^x }
p6eval niecza v13-332-g677cf15: OUTPUT«===SORRY!===␤␤Placeholder variable $^x may not be used here because the surrounding package block takes no signature at /tmp/Z5NeXH4TUF line 1:␤------> role a { ⏏$^x }␤␤Unhandled exception: Check failed␤␤ at /home/p6e… 12:05
jnthn moritz: Did you allow it for roles? 12:06
moritz jnthn: yes
jnthn hehe :) 12:07
moritz $*PKGDECL ne 'role' && ...
jnthn *nod*
lumi___ nom: class A is Hash { }; my $x = A.new; say $x.WHAT;
p6eval nom 63663b: OUTPUT«Hash()␤»
lumi___ That's not right, is it?
moritz it's not 12:08
method new(*args) { @args.hash }
jnthn wonders how on earth Hash's .new method has been written...
...ouch. 12:09
moritz perl6: class A { method x { $^x } } 12:10
p6eval rakudo 63663b: OUTPUT«===SORRY!===␤Placeholder variables cannot be used in a method at line 1, near "}"␤»
..pugs b927740, niecza v13-332-g677cf15: ( no output )
jnthn std: class A { method x { $^x } } 12:11
p6eval std 40d10f7: OUTPUT«ok 00:01 108m␤»
jnthn Hm. 12:12
moritz rakudo now takes about a quarter of an hour to compile the setting 12:37
which significantly slows down my development cycle
to be more precise, 17 minutes alone for the .pm -> .pir step 12:38
jnthn moritz: Is that a sudden jump?
moritz jnthn: dunno, it felt slow for some time, and now it feels really, really slow 12:39
jnthn moritz: OK. Will profile, but I strongly suspect the serialization work is going to be what we really need. 12:41
That's how it looked last time I did so.
Once we get a Star release out, I'll be happy to dig back into these big picture things.
moritz ok 12:42
... plus another 4.5 minutes for .pir -> .pbc
jnthn Virtually all of which is spent in register allocation. 12:44
I should be able to make a start on the serialization stuff within the next week. 12:47
I've got a bunch of the design work for it already sketched out in my notebook.
jnthn still wonders how on earth it takes so long and thinks something is scaling non-linearly. 12:48
12:50 fsergot left
flussence o/ 12:51
jnthn, seen this? doc.rust-lang.org/doc/tutorial.html...reign-code
the blockish syntax seems like a good idea, I'm wondering if there's an easy way to do something similar in perl6 code 12:53
13:02 kaleem left, kaleem joined 13:07 kaleem left
moritz wow, just found glx.sourceforge.net/index.html 13:08
13:12 avar left, literal left, groky left
jnthn flussence: Which "blockish syntax" are you refering do? 13:12
fn SHA1(src: *u8, sz: uint, out: *u8) -> *u8;
Would be something like
hmm, I dunno that languages syntax well enough to know what *8 is 13:13
moritz *u8 is just char*
jnthn Ah
moritz (in first approximation) 13:14
flussence no, I mean the "native mod crypto { ... }"
jnthn sub SHA1(Str, Int) returns Str is native('crypto') { } # or some such
flussence (as opposed to "is native()" on every sub
)
jnthn Aha
13:15 avar joined, avar left, avar joined, literal joined, groky joined
jnthn Interesting idea. 13:15
moritz nom: { $^a; $^a } 13:17
p6eval nom 3ddd2f: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in block <anon> at /tmp/bTWJkorzpl:1␤ in block <anon> at /tmp/bTWJkorzpl:1␤ in <anon> at /tmp/bTWJkorzpl:1␤»
moritz nom: do { @_ }
p6eval nom 3ddd2f: OUTPUT«===SORRY!===␤Cannot use placeholder parameters in this kind of block␤at /tmp/JoseEFE9NX:1␤»
flussence hm... I wonder if this could work 13:20
moritz flussence: push it onto the growing stack of stuff that masak++'s macro work should accomplish :-) 13:21
13:22 xinming joined
flussence oh, why didn't I think of using that! :) 13:22
flussence was trying to figure out a way to mess with the Routine object in a lexical scope, which probably wouldn't DWIW anyway 13:23
13:24 _twitch joined
[Coke] jnthn++ # starring 13:28
jnthn: do you anticipate any growls of "hey, I thought you said this was STABLE"?
jnthn: s/another month/another release/ ? 13:31
otherwise, all looks good to me!
13:36 cognominal_ left
[Coke] sorear: any plans for the last week before the next release? 13:38
13:39 drbean left 13:47 cognominal joined
dalek ast: 6c236a2 | coke++ | S29-context/exit.t:
niecza fudge
 13:57
ecza: a8940ad | coke++ | t/spectest.data:
Run (fudged) S29-context/exit.t
 13:58
kudo/nom: 2c3f642 | moritz++ | src/ (2 files):
include parameter name in error message

also more awesome error message stolen from STD and niecza		 13:59
ast: f52c2ed | moritz++ | S32-exceptions/misc.t:
test presence of placeholder attribute of X::Placeholder::Block exceptions
 14:00
jnthn [Coke]: People will always find something to growl about. :) I think if we'd done a Star distribution release on nom a month or two ago, there woulda been quite a few howls. Nowadays, I think much less so - the benefits of this release generally far outweigh the few regressions here or there. 14:01
[Coke]: Take MiniDBI. My changes were entirely to update it to use the latest NativeCall module and to fix places it did stuff that was no longer in compliance with current spec. No changes due to nom bugs. 14:02
lumi___ moritz: There's a typo in the message in your last commit: "may not be usere" 14:06
moritz uhm, the whole sentence reads as if I had been drunk :/ 14:08
lumi___ I stopped parsing there :P
dalek kudo/nom: bb2f072 | moritz++ | src/core/Exceptions.pm:
fix error message, lumi++
 14:09
kudo/nom: 355430c | moritz++ | src/core/Exceptions.pm:
another spelling fix, kboga++ moritz--
 14:10
ast: 6b9e653 | coke++ | S12-methods/multi.t:
fudge for niecza
 14:11
ecza: 9736793 | coke++ | t/spectest.data:
run (fudged) S12-methods/multi.t
14:15 kaleem joined
tadzik I thought "usere" is the awesomeness in this message :) 14:16
usere as in sincere, or something :)
14:17 Alias joined
[Coke] checks for LowHangingFudge in Rakudo. 14:20
moritz [Coke]: autounfudge? 14:21
[Coke] ugh. probably 100 more. 14:22
moritz: no, update_passing...
moritz ah
[Coke] one with 47 passing, another with 20, lots of little ones. 14:23
14:24 Exodist left 14:25 jjore left 14:26 Exodist joined 14:27 jjore joined
jnthn wonders if S05-mass/rx.t wouldn't have a bunch more after the fixed backtracking into subrules 14:28
moritz jnthn: possible. I'll run autounfudge once my current build (on newest parrot) finishes 14:29
[Coke] moritz: S24-testing/0-compile.t passes 100% as is.
moritz [Coke]: I've never enabled that because it contains only stuff we test elsewhere 14:30
[Coke] and t/spec/S32-hash/kv.t is fudged, but not being run.
moritz [Coke]: and the whole S24-testing looks suspicious to me
[Coke] moritz: if the spec test is invalid, we should remove it. *shrug*
moritz it's not invalid. It just doesn't make much sense 14:31
[Coke] ... if it doesn't make sense, why are we keeping in the repository and forcing people to think about it?
invalid doesn't necessarily mean "not to spec." 14:32
[Coke] now remembers why he doesn't do this for rakudo. ;)
moritz I've simply never mustered the energy to check S24-*/*.t more closely. If you want to remve it, go ahead 14:33
jnthn [Coke]: ? 14:36
[Coke] jnthn: ? ? 14:38
jnthn [Coke]: Why you don't do what for Rakudo?
[Coke] identify more tests that can be run. 14:39
though it was a snarky comment, in that I don't actually remember this as a problem.
14:39 PacoAir joined
[Coke] feather.perl6.nl/~coke/rakudo_update_passing.txt has a fresh run of update_passing_test_data.pl 14:41
moritz [Coke]: thanks 14:42
14:42 Trashlord joined, renormalist joined
[Coke] (S05-modifier was nontrivial to fudge) 14:42
er, feather.perl6.nl/~coke/rakudo_update_passing.txt
er, t/spec/S05-modifier/counted.t !
jnthn looks confusedly at the output 14:43
[ PA] ( 0/1 ) t/spec/S01-perl-5-integration/import.t
[ A] = all passed
...why does that get A? :)
moritz jnthn: maybe one was skipped (and thus passed) 14:44
jnthn ah, ok
moritz jnthn: or maybe something else. The harness doesn't give us all the numbers that I want/need, so I calculate the rest; maybe some of them are a bit off
[Coke] i just use it as a guidline for unrun tests to look at, higher numbers generally being the first things to touch. 14:45
(though anything with SPA is usually an easy win.)
moritz [Coke]: that was my original intention when I wrote that script 14:46
so it works as designed, even if not always free of surprises :-) 14:47
14:51 cognominal_ joined 14:54 cognominal left 14:56 renormalist left 15:08 fsergot joined 15:09 birdwindupbird left 15:11 pmurias joined
pmurias sorear: hi 15:11
sorear: i'm thinking about implementing use Foo:from<perl5>
sorear: the hard parts seems to be that it happens at compile time but we also need to load stuff at runtime 15:13
moritz but not with 'use'
pmurias sorear: forget the last sentence ;) 15:15
sorear: the hard part is that use loads stuff into the lexical namespace at compile time, but we want to use the perl5 things at runtime 15:16
sorear: and i'm not aware of a sane way to serialize code refs
pmichaud good morning, #perl6
moritz \o/ pmichaud! 15:17
pmurias pmichaud: good morning
15:17 tokuhirom joined
pmurias sorear: and p5 modules don't have a clear runtime/compile time distinction so they may depend on it being in the same process 15:17
sorear: what seems to be the best solution seems load stuff into the lexical namespace as placeholders, that is &foo would know it's imported from Foo 15:18
[Coke] pmichaud: Good morning!
pmurias sorear: and then replace them at runtime 15:19
[Coke] for a second, I thought moritz had mis-tabbed again. ;)
tadzik hello pmichaud
moritz colomon: as if I'd ever mis-tab :-) 15:21
pmichaud [Coke]: yes, it's really me.
moritz: :-P
pmurias moritz: re "not with use" ?
jnthn pmichaud!! \o/ 15:22
moritz pmurias: "use" + p5 doesn't do any more runtime loading than "use" + p6
15:23 arlinius left
jnthn
.oO( Just as predicted, pmichaud shows up within 48 hours of me fixing backtracking into subrules :P )
 15:23
moritz pmurias: when precompiling a module that loads a p5 module, you'll not get around loading the p5 module twice 15:24
because use Foo:from<perl5>; BEGIN run_imported_sub(); should work
jnthn: so what was the plan? let him fix all the list issues? :-)
15:24 arlinius joined
jnthn moritz: I suspect that depends on the tuit supply. ;-) 15:25
pmichaud jnthn: :-)
my tuit supply should be pretty good. today is a little busy, but I should get some good hacking in this week
moritz \o/
jnthn pmichaud: yay! 15:26
tadzik wow, great!
jnthn will have lots of availability this week too :)
pmurias moritz: the difference between "use"+p5 and "use"+p6 is that we can serialize the stuff from the p6 module much more easily (as we control the implementation) 15:30
re 'BEGIN run_imported_sub()' i could sacrifice that but even discovering what &import imports requires loading the p5 module 15:32
15:33 _twitch left
[Coke] jnthn: ah, but now he can work on stuff that's actually hard. ;) 15:37
dalek ast: 40f4b64 | moritz++ | S03-operators/range.t:
rakudo unfudges for non-numeric ranges
 15:38
15:38 Chillance joined 15:39 pmurias_ joined
jnthn [Coke]: :P 15:40
moritz jnthn: the autounfudge run didn't find any newly passing regex tests, but that might be because the resource limit is a bit too tight 15:41
15:41 pmurias left
moritz jnthn: but it could also be that S05-mass/rx.t doesn't use backtracking over captures a lot, because it doesn't use many captures at all 15:42
jnthn moritz: ah, k 15:43
15:44 thou joined
moritz perl6: say "\x5" 15:44
p6eval pugs b927740, rakudo 355430, niecza v13-332-g677cf15: OUTPUT«␤»
moritz perl6: say "\x5c"
p6eval pugs b927740, rakudo 355430, niecza v13-332-g677cf15: OUTPUT«\␤»
15:44 GlitchMr left 15:45 GlitchMr joined, wtw_ left 15:47 GlitchMr left, GlitchMr joined 15:48 jferrero left
jnthn decommute & 15:49
dalek ast: e10ffa7 | moritz++ | S05-mass/rx.t:
correct two tests
 15:50
moritz niecza might pass tests too
nom: say 'a\\b'
p6eval nom 355430: OUTPUT«a\b␤»
moritz nom: say 'a\\b' ~~ /"a\\b"/
p6eval nom 355430: OUTPUT«=> <a\b>␤␤»
moritz nom: say 'a\\b' ~~ /"a\b"/ 15:51
p6eval nom 355430: OUTPUT«#<failed match>␤»
15:56 not_gerd joined
not_gerd hello, #perl6 15:57
moritz \o not_gerd 15:58
not_gerd [Amoritz: could you pull github.com/parrot/parrot/pull/704 into a new branch in the Parrot repo (NOT master)?
(no one else around in #parrot...)
16:00 zb left
moritz not_gerd: doing so now 16:00
16:01 nwc10 joined
nwc10 Yay, a pmichaud! 16:01
moritz not_gerd: pushed to parrot/auto-sizes 16:03
not_gerd moritz: thanks
16:04 kaare_ joined 16:05 SHODAN left 16:06 SHODAN joined
not_gerd jnthn is the only one around with Win64 + MSVC, correct? 16:07
[Coke] aye. I have win32 + msvc, but don't often build on it. 16:09
16:10 simcop2387 left, simcop2387_ joined 16:11 simcop2387_ is now known as simcop2387 16:13 Trashlord left
jnthn home 16:24
not_gerd: Yes, I build very often on that.
not_gerd: Does that fix the need to do --intval="long long" on Win64? 16:25
not_gerd jnthn: yes, among other things 16:28
jnthn not_gerd: Excellent.
not_gerd still accepts standard-integer types only, though, as auto::format will fail if you do something like --intval=__int64 16:29
[Coke] pmichaud: mind if I ask what's on your TODO list this week? Any itches need scratching? 16:31
pmichaud you mean perl6-wise? my todo list this week is to catch up and see where things stand and what needs doing
lists sounds like the prime target (as well as the corresponding synopsis/grant) 16:32
jnthn pmichaud: Yes, lists decidedly need some love.
pmichaud I suspect jnthn++ would like me to double-check the regex engine stuff too :)
jnthn pmichaud: Please. ;) 16:33
16:34 kaleem left
not_gerd jnthn: if you have a minute, could you check that NQP's --gen-parrot=auto-sizes works out-of-the-box on Win64? 16:37
jnthn not_gerd: Yes, I shall do that shortly.
not_gerd thanks
16:37 nwc10 left 16:43 pmurias_ left 16:44 kmwallio joined 16:45 colomon joined 16:51 fsergot left 16:53 rindolf joined 16:55 alester joined 17:01 tokuhirom left 17:02 kmwallio left 17:07 jdv79_ is now known as jdv79 17:08 jdv79 left, cognominal joined 17:09 MayDaniel joined
jnthn not_gerd: Fails :( 17:10
.\miniparrot.exe -Iruntime/parrot/include -o frontend/parrot2/prt0.pbc frontend/parrot2/prt0.pir
PackFile_pack segment 'DIRECTORY' used size 2368 but reported 2664
17:11 cognominal_ left
rindolf Hi all. 17:12
not_gerd jnthn: :( 17:13
17:14 fsergot joined
not_gerd first step: could you diff a config_lib.pir from a working and a non-working parrot? 17:14
if there's no difference, then I messed up when doing the unrelated cleanup... 17:16
masak evenin', #perl6 17:21
jnthn ohhai, masak
moritz \masak/
masak \:D/
17:22 pmade joined
jnthn not_gerd: gist.github.com/1664386 17:29
colomon phenny: tell sorear in S14-roles/mixin.t, $x does R4b(3) test is failing. 17:30
phenny colomon: I'll pass that on when sorear is around.
not_gerd jnthn: possible suspect is the change in opcodesize, but I'll probably need help from the parrot folks to debug that... 17:33
17:35 mj41 left
not_gerd jnthn: I think I know what's going on and can fix it on my own... 17:36
17:39 Trashlord joined 17:42 fsergot left
not_gerd jnthn: gist.github.com/1664459 _should_ do the trick... 17:44
17:47 Radvendii joined
Radvendii o/ 17:47
How do I run a map over key-value pairs? if i do @a.kv.map: {} It takes the keys and values as separate input. How do i make them get passed to .map at the same time 17:49
jnthn Radvendii: .pairs
Radvendii: Or .kv and have your map block take two args
Radvendii jnthn: sweet. thanks
TimToady or map ->$k,$v {...}
Radvendii ooh. i didn't know you could do that
cool
TimToady oh, jnthn++ said it 1st
jnthn ;) 17:50
Radvendii thanks guys
TimToady ENOCOFFEEYET
jnthn not_gerd: Trying a build with that now 17:53
Radvendii it keeps saying Can only use repr_get_attr_obj on a SixModelObject. this arises from initializing a hash instance variable from an array instance variable
what am i doing wrong?
jnthn Radvendii: Hard to say without seeing your code 17:54
Radvendii never mind
fixed
i forgot to comment out some code
jnthn not_gerd: Seems that helps :) 17:55
17:55 Trashlord left
not_gerd jnthn: great 17:55
now there's a question: why do we have separate configure options for intval and opcode if things break if we provide different values? 17:56
17:57 kaleem joined
jnthn not_gerd: That's one you'll probably have to ask on #parrot ;) 17:58
moritz hysterical raisins 17:59
somebody thought at some point it was necessary to be able to configure both, nobody ever tested it with diverging values, and thus it's broken
that's how it always goes
jnthn OK, shopping, dinner, then I'll have some hacking tuits 18:00
bbl
18:00 dakkar left
sorear good * #perl6 18:00
phenny sorear: 17:30Z <colomon> tell sorear in S14-roles/mixin.t, $x does R4b(3) test is failing.
Radvendii o/ jnthn
18:00 Trashlord joined
colomon \o 18:00
sorear [Coke]: The last big thing I want to squeeze in is MAIN 18:01
18:03 jaldhar left
[Coke] sorear: as a coding contest contestant, that'd be nice! 18:06
18:10 MayDaniel left
Radvendii is there an equivalent in perl6 to overloading perl5 STORE? 18:10
as in, a method that get's called when i do $a<key> = 5; as different from when i simply access $a<key> 18:13
sorear No, it doesn't work like that
$a<key> = 5 calls the same method that $a<key> does 18:14
$a<key> = 5 means $a.at-key('key') = 5
make sure at-key returns something that can be assigned to
dalek ast: 52e9965 | sorear++ | S14-roles/mixin.t:
[S14-roles/mixin] Do not attempt to mutate Int objects here, either
 18:15
Radvendii so if i overload at-key (is that the same as at_key?) i can't know what is being assigned to it?
tadzik maybe the Proxy object is what you want 18:16
18:16 fsergot joined
Radvendii Proxy? 18:16
there seems to be a STORE_AT_POS method of Array 18:18
tadzik see perlcabal.org/syn/S06.html#Lvalue_subroutines
pmichaud STORE_AT_POS is typically used for creating new array elements 18:21
TimToady o/ 18:22
sorear Radvendii: I do not think you understand the core principle at work here
pmichaud TimToady: o/
Radvendii probably not...
sorear Radvendii: '=' does not care what is on the left
TimToady (unless it's pseudoassignment) 18:23
18:23 flussence left
sorear Radvendii: every expression in Perl 6 can return a 'container', which is something that can be either read or written 18:23
18:23 flussence joined
sorear PMICHAUD! \o/ 18:23
18:24 _RobiX_ joined
pmichaud sorear: o/ 18:25
sorear TimToady: didn't you just get rid of pseudoassignments? 18:27
TimToady no, I just turned assigments into pseudoassignments
[Coke] ponders doing another spectest run before pmichaud starts adding things. ;)
18:27 _RobiX_ is now known as hello2u2 18:28 xinming left
TimToady the = is now parsed by the declarator, not by EXPR 18:28
(but only for declarators, of course--normal assignment is still, er, normal) 18:29
18:31 spine joined
sorear [Coke]: I only fixed one test ... 18:32
18:33 pmade left
PerlJam wonders if the "different things should look different" rule will ever kick in. 18:33
TimToady what would you suggest that will not provoke widespread disdain? :) 18:34
contextually, it's pretty easy to distinguish from normal assignment for a human, I think 18:35
we could go all cobolly and say 'my $x INITIALIZED TO 42' :P 18:37
jnthn Radvendii: Were you not delegating to a hash/array that you "has"? If so, it returns a container - just be sure to mark your at_pos with "is rw", and either let the result be the last statement or use return-rw instead of return on it.
std: sub infix:<INITIALIZED TO>($a, $b) { }; my $x INITIALIZED TO 42
p6eval std 40d10f7: OUTPUT«Potential difficulties:␤ $b is declared but not used at /tmp/esTkeEiMbH line 1:␤------> sub infix:<INITIALIZED TO>($a, ⏏$b) { }; my $x INITIALIZED TO 42␤ $a is declared but not used at /tmp/esTkeEiMbH line 1:␤------> sub infix:<INITIALIZED T…
PerlJam TimToady: I like nice, simple rules. Telling people that = is assignment *except* when it comes after a declarator isn't so simple.
TimToady but = has the advantage of being very easy to type too 18:38
jnthn std: sub infix:<INITIALIZED TO>($, $) { }; my $x INITIALIZED TO 42
p6eval std 40d10f7: OUTPUT«ok 00:01 116m␤»
[Coke] sorear: there's the 13 or so you fixed yesterday. I added a dozen or two.
jnthn heh :)
TimToady same reason C stole it from ==
18:40 rindolf left
TimToady PerlJam: human languages are most irregular at the point of greatest use, if you want to view this as an irregularity 18:40
flussence
.oO( we could steal "my $x default 42" from SQL, but that's not much better than COBOL... )
 18:41
PerlJam flussence: *way* too much typing
flussence agreed :)
TimToady
.oO(use COBOL :lc;)
PerlJam my $x is (42); # crazy? 18:42
maybe never mind the parens, I'm not sure where I was going with them.
TimToady hard to get a list into it like my @a = 1,2,3 18:43
at least in sigs, 'is' is tighter than comma
with pseudoassignment we can at least preserve the $ vs @ distinction on precedence of = 18:44
18:44 hello2u2 left
moritz as a user, I like the pseudoassignment initialization 18:44
TimToady it's kind of a useful confusion :)
moritz as a compiler writer, I like it as long as somebody else is responsible for it :-)
flussence
.oO( let's get rid of `has` entirely and give classes signatures like subs! :)
 18:45
PerlJam maybe pseudoassignment bothers me so much because I foresee another FAQ entry ... but maybe some FAQs are just a necessary evil
dalek kudo/nom: d469892 | moritz++ | t/spectest.data:
enable kv.t, [Coke]++
18:46 hello2u2 joined, hello2u2 left, rindolf joined
moritz PerlJam: at least we now know where to put them. faq.perl6.org :-9 18:46
sorear o/ rindolf
rindolf sorear: hi. 18:47
sorear: I got disconnected.
TimToady if we wanted a bit more regularity, we could actually take back the precedence-dropping of pseudoassignments into normal assignments, so all list assignments parse like listops
PerlJam "How do I overload assignment and pseudoassignment?" or perhaps "I've implemented my own assignment operator, but it doesn't appear to work with variable declarations.."
TimToady but I don't know that we need to do that 18:48
moritz PerlJam: the answer is probably "don't mess with assignment" :-)
PerlJam Perl's answer is never "don't"
TimToady Perl's answer is: "Easy, write your own compiler!"
PerlJam It's always "proceed at your own risk and watch out for the pointy bits!" 18:49
moritz ok, maybe the answer should be "If you have to ask, you're not ready to overload assignment"
PerlJam moritz: perhaps
TimToady It could be a very short FAQ entry: "No, for certain values of no." 18:52
moritz TimToady: an MJD-style FAQ :-)
TimToady invents the Frequently Unanswered Questions ;) 18:53
PerlJam heh
TimToady bows 18:54
sorear just so long as we don't get a FQA...
dalek q: 4aa6be5 | moritz++ | answers.md:
"Why sigils?"

Mostly compiled from irclog.perlgeek.de/perl6/2012-01-19#i_5016530
18:54 PacoAir left
moritz sorear: we just give potential FQA answers commit access to faq.perl6.org :-) 18:54
[Coke] NOMMAP still hints that nom will become master someday. 18:55
TimToady just so we don't get a FQU too
PerlJam [Coke]: it still might!
masak I gave a Perl 5 course today. the question wasn't so much "why sigils?" as "why variant sigils?"
PerlJam :-)
masak (from the attendees) 18:56
TimToady former Perl 6 programmers? :)
dalek kudo/nom: 268e0c9 | moritz++ | NOMMAP.markdown:
nom will not become master. [Coke]++
 18:57
masak TimToady: probably in a prior life or sump'n.
[Coke] nom: my @a; say @a.WHAT; 18:59
p6eval nom 355430: OUTPUT«Array()␤»
TimToady ah well, not suprisingly, looks like other people invented FUQ first
masak .oO( aw FUQ ) 19:00
19:00 thou left
TimToady FQU too 19:00
[Coke] moritz: I don't know how to ask @a if it's the type object. If it is, that covers one of the remaining TODOs.
sorear @a has to be defined 19:02
unless you beleive the stuff in S02 about assigning Nil to aggregate containers, but that doesn't make much sense
pmichaud [Coke]: I'm not likely do be doing any significant committing until tomorrow or a bit later in the week 19:03
it may take me that long to get re-settled in here
PerlJam rakudo: my @a := Nil;
p6eval rakudo d46989: OUTPUT«Type check failed in binding␤ in block <anon> at /tmp/7Xz5wuXvnv:1␤ in <anon> at /tmp/7Xz5wuXvnv:1␤»
PerlJam rakudo: my @a := ();
p6eval rakudo d46989: ( no output )
jnthn nom: say Nil ~~ Positional
p6eval nom d46989: OUTPUT«Bool::False␤»
pmichaud plus I still have a lot of cleaning up to do at the house
PerlJam Shouldn't those be the same? Or am I outdated in my thinking?
masak pmichaud! \o/ 19:04
pmichaud masak: o/
masak \o/ \o/ \o/
moritz [Coke]: @a.defined
perl6: my @a; say @a.defined
p6eval rakudo d46989, niecza v13-332-g677cf15: OUTPUT«Bool::True␤»
19:04 crashtest joined
p6eval ..pugs b927740: OUTPUT«1␤» 19:04
masak crashtest: aloha.
sorear PerlJam: I'm talking about my @a = Nil, not :=
pmichaud rakudo; my @a; say @a.defined; # does Rakudo have this correct yet? 19:05
PerlJam sorear: I'm aware. You're assertion led me to muck with := though
pmichaud rakudo: my @a; say @a.defined; # does Rakudo have this correct yet?
p6eval rakudo d46989: OUTPUT«Bool::True␤»
PerlJam s/You're/Your/
pmichaud no, it doesn't.
benabik nom: my @a; say @a.perl 19:06
sorear pmichaud: as far as I'm concerned that *is* correct
p6eval nom d46989: OUTPUT«Array.new()␤»
masak pmichaud: me too.
pmichaud: spec?
sorear b: my @a; say @a.defined
moritz so should it autovivify to a defined Array on assignment?
p6eval b 1b7dd1: OUTPUT«Bool::True␤»
sorear look, all four major implementations agree, @a makes a defined Array
masak I like that it's defined.
(I think.)
moritz masak: I think you were one of the proponents for having it undefined 19:07
pmichaud I think we've had discussions on this in the past; also, p5 says that uninitialized arrays are undefined
masak moritz: yeah, that's what makes me uncertain.
moritz so that one could distinguish sub f(@a?) { } f(()); f();
masak moritz: I remember the discussions.
sorear eval: my @foo; defined @foo
buubot_backup sorear: No output.
sorear eval: my @foo = (); defined @foo
buubot_backup sorear: No output.
masak moritz: not so sure anymore I was right.
sorear eval: my @foo = (3); defined @foo
buubot_backup sorear: 1
sorear eval: my @foo = (3); @foo = (); defined @foo
buubot_backup sorear: 1
moritz masak: I guess nowadays you don't often check if optional arguments have been filled 19:08
masak moritz: whether something was passed or not is more a property of the Capture than of the variable in the sub.
moritz masak: because you've (maybe subconciously) realized it's often design smell
that too
masak moritz: it feels... wrong to laden the variable with that information.
Radvendii @people, thanks for the help (way up there). The issue is that I want to have the hash and the array linked, so that when you modify the array part, it still modifies the array part, but also does something to the hash part depending on what is being assigned
sorear pmichaud: looks like p5's rule is actually "has this array ever held an element", not "has this array been initialized" 19:09
masak Radvendii: that's entirely possible. just implement your own class with custom .[] and .{}
pmichaud sorear: okay, good correction.
masak Radvendii: I could put together a small example for you if you want.
sorear Use of "defined" on aggregates (hashes and arrays) is deprecated. It used to report whether memory for that aggregate has ever been allocated. This behavior may disappear in future versions
of Perl. You should instead use a simple test for size:
does Perl 6 count as a future version of Perl? 19:10
masak Radvendii: sort of like a sorted hash or something like that?
moritz ... and it would be crazy to replicate p5s behaviour in p6
19:10 thou joined
Radvendii masak: I tried, but [] is unaware of what value is being "equalled" to it. 19:10
19:10 thou left, thou joined
pmichaud fwiw, the "all four implementations agree" just means that all four could be wrong. We're looking at Second System Syndrome Done Right, even if all of the implementations are Wrong. :-) 19:11
masak Radvendii: I did not understand that response. we might be talking at cross purposes.
19:11 DarthGandalf left
masak pmichaud: you have a point -- but that doesn't mean you're right :P 19:11
TimToady p5's behavior is insane because it depends on the allocatedness of its memory, not on the actual semantics in question
pmichaud I don't have a strong opinion as to whether "my @a" should be defined or undefined; I just know that it's been declared differently from what Rakudo does in the past.
rephrase: I just know that in the past it's been declared to be different from what Rakudo is doing now. 19:12
Radvendii masak: When i say @a[0] = 1, it calls @a.[] but if i overload .[], it can't know that 1 is being assigned to it, which is what i need.
TimToady I think it's very useful for any bindable variable to have a state that says it is not meaningful
masak pmichaud: well, I think that declaration is part of a long rambling open RT ticket that someone named masak (no relation) filed once.
moritz Radvendii: you can return a Proxy object from @a[0], but it is usually a sign of design smell if you need to do that 19:13
masak Radvendii: maybe you need to overload at_pos?
sorear pmichaud: hey, you just got back and I'm already getting to the antagonizing-you stage. Sorry :|
masak moritz: what? Proxy objects are a design smell now!? :)
Radvendii masak: I've been making the problem more abstract, what I am doing is making a chess board which is represented both as a two-dimensional array of spaces, and as a hash of pieces (and where they are on the board)
benabik moritz: I'm not so sure it's design smell. Being able to enforce invariants at assignment is useful.
masak sorear++ # tormenting implementor
sorear pmichaud: welcome back. I hope thinks are going adequately well for you
pmichaud sorear: no problem; I've missed that a bit. :-) 19:14
moritz masak: usually.
Radvendii masak: i am overwriting at_pos
masak moritz: I thought Proxies were the best.
moritz nom: my Int $x = 3; sub f() is rw { $x }; f() = 'foo'
pmichaud and yes, things are going adequately well here.
p6eval nom d46989: OUTPUT«Type check failed in assignment to '$x'; expected 'Int' but got 'Str'␤ in block <anon> at /tmp/xkuqahwsvL:1␤ in <anon> at /tmp/xkuqahwsvL:1␤»
19:14 preflex left
moritz benabik: that also enforces an invariant at assignment, and doesn't need a proxy 19:14
Radvendii oh oops. i've been putting the wrong name at the beginning of my messages 19:15
jnthn nom: my Int $x = 3; sub f() is rw { return-rw $x }; f() = 'foo'
p6eval nom d46989: OUTPUT«Type check failed in assignment to '$x'; expected 'Int' but got 'Str'␤ in block <anon> at /tmp/UqhksAIsX2:1␤ in <anon> at /tmp/UqhksAIsX2:1␤»
PerlJam Radvendii: that's okay, we always get them confused too ;)
jnthn oh, duh, I didn't read the error :)
TimToady doesn't channel MJD
benabik moritz: A language invariant, yes... But there are invariants that can't be as easily enforced that way... Like "array can't contain duplicates" 19:16
PerlJam
.oO( * TimToady doesn't channel MJD, RETARDO!? )
masak moritz: I think in order to claim that something like a Proxy is a design smell, it'd be nice if you could say why you think so. if you said roles are a design smell, I wouldn't believe you. with Proxy, I don't know.
pmichaud I have to run errands now; unlikely I'll be back much more today. I should be around most of tomorrow, however.
masak PerlJam: :D 19:17
[Coke] pmichaud: be well.
Radvendii pmichaud: o/
masak pmichaud: o/
moritz masak: my statement might have been a bit harsh; what I should have said is that IME people tend to reach for proxies when there are better options, just like people overuse junctions 19:18
19:18 GlitchMr left 19:19 DarthGandalf joined, dbr left
Radvendii In any case, I am using at_pos. The problem is at_pos is unaware of what is being assigned, and it is called before the assignment takes place (so i can't even do an update of the entire hash. 19:19
sorear niecza: class A { method foo { say @_; say self } }; A.foo(5) 19:20
p6eval niecza v13-332-g677cf15: OUTPUT«5␤A()␤»
moritz Radvendii: then a Proxy is indeed what you need
jnthn Radvendii: It's meant to work like that.
Radvendii i suppose i could just represent it as an array, and convert to a hash every time they ask for the value at a key... 19:21
moritz: how would a Proxy help?
moritz Radvendii: it lets you run a piece of code when somebody assigns to it
PerlJam Radvendii: it's too bad shaped array aren't implemented anywhere.
Radvendii PerlJam: shaped arrays?
moritz see S09
Radvendii moritz: oh! how do i do that?
ah
moritz Radvendii: there are examples in S06 and in the test suite 19:22
just grep for Proxy
sorear out
masak moritz: ok -- to me "code smell" and "cargo-cult overuse" are two different mental bins. 19:23
PerlJam Radvendii: I guess I really mean "user-defined array indices", but I tend to think of them in conjunction with shaped arrays.
moritz masak: for me too, but I often conflate them when talking to somebody who doesn't express where well what he wants
masak it's hard to know what you want in a language you're in the process of learning :) 19:24
moritz masak: aye 19:25
Radvendii PerlJam: I don't understand what you mean...
jnthn nom: class A { has @.x; method at_pos($self: $i) is rw { Proxy.new: FETCH => method () { $self.x[$i] }, STORE => method ($val) { $self.x[$i] = 2 * $val } } }; my $a = A.new; $a[0] = 21; say $a[0]
p6eval nom d46989: OUTPUT«42␤»
jnthn Radvendii: ^^ example
Radvendii jnthn: yes. thanks. i just found the part in 06. 19:26
19:26 birdwindupbird joined
PerlJam Radvendii: you can safely ignore me, what I was thinking of would only satisfy a part of what you want. 19:26
19:26 dbr joined
Radvendii do i have to impliment anything but FETCH and STORE? 19:26
masak nom: say ('bird' xx 2).join('windup')
p6eval nom d46989: OUTPUT«birdwindupbird␤»
[Coke] b: say 19284-19269; # 01/23/2012 - niecza at 99.92%
p6eval b 1b7dd1: OUTPUT«15␤»
Radvendii PerlJam: okay... I'll look it up later. it does look interesting
jnthn Radvendii: Proxy only knows about FETCH and STORE :) 19:27
[Coke] niecza still has 10 fails, rakudo 28.
jnthn [Coke]: overflow.t?
Radvendii jnthn: awesome.
masak loves living in a post-Easter world
moritz
.oO( now where are my Perl 6 Eaters? )
 19:28
[Coke] feather.perl6.nl/~coke/ has the individual reports if you want to dig for the failures.
jnthn: yup
jnthn [Coke]: Thanks
[Coke] niecxza is still the TERMCAP issues.
ugh, 15 tests. :P 19:29
19:29 birdwind1pbird joined, cognominal_ joined 19:32 cognominal left
dalek ast: 22fb63e | jnthn++ | integration/99problems-31-to-40.t:
Bring a test inline with spec.
 19:32
19:32 kaleem left 19:34 daxim left
dalek ast: 3dd2b4d | coke++ | integration/99problems-31-to-40.t:
unfudge for niecza
 19:35
19:35 sromanov joined
[Coke] treads water. ;) 19:36
moritz "I pass that test" -- "me too!"
:-)
TimToady I wish there were a niecza user so that I could autocomplete its name...
moritz should I rename hugme to niecza? :-)
[Coke] niecza: my @a = 4,6,12 .. *; say @a[3]; 19:37
TimToady nah, rename buubot if you really want to confuse people
p6eval niecza v13-332-g677cf15: OUTPUT«(timeout)»
TimToady .. is not ...
though that shoulda worked, I guess 19:38
on the theory that 12 .. * should transmute to 12 ... *
[Coke] niecza: my @a = 4,6,12...*; say @a[^10];
p6eval niecza v13-332-g677cf15: OUTPUT«Unhandled exception: Unable to figure out pattern of series␤ at /home/p6eval/niecza/lib/CORE.setting line 1168 (die @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2555 (get-next-closure @ 25) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2567 (AN…
TimToady well, it's that my's pseudoassignment is eager
[Coke] so much for using perl 6 on this week's car talk puzzler. ;) 19:39
TimToady unlike other pseudoassignments
[Coke]: you're looking for constant, I think
moritz niecza: my @a := 4, 6, 12 ... *; say @a[^10]
p6eval niecza v13-332-g677cf15: OUTPUT«Unhandled exception: Unable to figure out pattern of series␤ at /home/p6eval/niecza/lib/CORE.setting line 1168 (die @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2555 (get-next-closure @ 25) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2567 (AN…
TimToady niecza: constant my @a = 4,6,12...*; say @a[^10];
p6eval niecza v13-332-g677cf15: OUTPUT«===SORRY!===␤␤Malformed constant at /tmp/iI8A3FsHb7 line 1:␤------> constant my ⏏@a = 4,6,12...*; say @a[^10];␤␤Parse failed␤␤»
TimToady niecza: constant @a = 4,6,12...*; say @a[^10];
p6eval niecza v13-332-g677cf15: OUTPUT«Unhandled exception: Unable to figure out pattern of series␤ at /home/p6eval/niecza/lib/CORE.setting line 1168 (die @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2555 (get-next-closure @ 25) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2567 (AN…
TimToady niecza: constant @a = 4,6,12..*; say @a[^10];
p6eval niecza v13-332-g677cf15: OUTPUT«4 6 12..Inf Any() Any() Any() Any() Any() Any() Any()␤» 19:40
colomon 4,6,12 isn't a p6-solvable series, is it?
TimToady uh, ranges should flatten in list context
moritz colomon: nope
TimToady the ... never sees the 4 or 6 if you start with ..
dalek ast: 70ffbf0 | jnthn++ | integration/99problems-11-to-20.t:
Correction, re-fudging (remove some skips and turn some others into todos).
TimToady since .. is tighter than , 19:41
moritz jnthn: are you doing an epic walk through the fudges?
TimToady niecza: my @a; @a = 1..12; say +@a 19:43
p6eval niecza v13-332-g677cf15: OUTPUT«12␤»
TimToady niecza: constant @a = 1..12; say +@a
p6eval niecza v13-332-g677cf15: OUTPUT«12␤»
jnthn moritz: More like just a gentle stroll :)
TimToady niecza: constant @a = 0,1..12; say +@a
p6eval niecza v13-332-g677cf15: OUTPUT«2␤»
TimToady Oopsie!
jnthn moritz: If you fancy what looks like 17 more easy passes, see S32-num/stringify.rakudo :) 19:44
moritz jnthn: you mean the Rat.perl thing in there?
TimToady sorear: ^^
jnthn moritz: Complex.perl also 19:45
moritz jnthn: problem is that it tests for a form that we don't parse as Complex
jnthn: so while we win tests, we lose roundtripping
jnthn moritz: ah.
19:45 Bzek joined
dalek ast: dca332d | moritz++ | S32-exceptions/misc.t:
test for typed redeclaration error
 19:46
kudo/nom: d83a1c0 | moritz++ | src/ (2 files):
typed Redeclaration errors
TimToady niecza: my @a = 0,1..12; say +@a 19:47
p6eval niecza v13-332-g677cf15: OUTPUT«13␤»
masak it's interesting that there's +/- and ?/!, but that prefix:<~> has no negation.
Radvendii when i do "@a[0][1]=1" it really does [email@hidden.address] correct? and the at_pos(0) would be using the FETCH part of the proxy, no? 19:48
TimToady perl6: say ~^ "FOO"
p6eval niecza v13-332-g677cf15: OUTPUT«Unhandled exception: Buffer bitops NYI␤ at /home/p6eval/niecza/lib/CORE.setting line 1168 (die @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2680 (prefix:<~^> @ 3) ␤ at /tmp/qhEnln36Dg line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.sett…
..rakudo 268e0c: OUTPUT«Failure.new(exception => Exception.new())␤»
..pugs b927740: OUTPUT«Â¹Â°Â°␤»
benabik Radvendii: The .at_pos(1) would use FETCH from the result of .at_pos(0)
Radvendii: Then the = 1 would use STORE on the one from at_pos(1)
Radvendii okay
that's good 19:49
thanks
19:49 Vlavv` left
TimToady Radvendii: the point is that we make a very careful distinction between determining the identity of the operand from performing the operation 19:49
subscripting is only for determining identity
Radvendii right. that makes sense. 19:50
TimToady likewise an rw function only identifies the operand, it never actually does anything to it
benabik @a[0][1] = 1 is approx. @a.at_pos(0).FETCH.at_pos(1).STORE(1) as I understand it.
Radvendii so how would i write a nested Proxy? Proxy.new: FETCH => method { Proxy/new: FETCH => method {}}? 19:51
but then where do i input the second index?
moritz Radvendii: you don't special-case nested access
benabik Radvendii: The first proxy returns something that has at_pos. 19:52
moritz Radvendii: if you want to allow @a[0][1], then @a[0] needs to return a Proxy which holds another object, which again can have a Proxy
19:52 GlitchMr joined
moritz if they are arrays, @a[0] simply returns an array 19:52
and that array knows how to deal with .[1]
Radvendii so i have to make a new class for my second level arrays? 19:53
moritz no
you can just reuse your custom array class
Radvendii then how do i overload their at_pos? 19:54
huh?
but the one that i am making also acts as a hash, it wouldn't work for this...
moritz Radvendii: are you familiar with recursion? 19:55
Radvendii yeah
o-oh
moritz just like with recursion, you just deal with one level
TimToady moritz: it's possible he wants a different type at the next level...
moritz and for the second level, you can return an object of the same type, which will handle the second level for you
TimToady: then it must be different types, yes 19:56
TimToady possibly with the same role mixed in though
Radvendii yeah... the top level is just returning the second level. Whereas the second level is doing weird things when you STORE to it
so i need another class?
TimToady sounds like a Node vs Leaf type problem 19:57
Radvendii can i do an inline overloading?
@a.overload : at_pos => method {} 19:58
?
moritz @a does role { method at_pos { } } 19:59
19:59 sromanov left
masak perl6++ 19:59
Radvendii okay... ("role" is kind of a misnomer)
moritz is it? it's just much more general than that particular use case 20:00
masak Radvendii: it's just an anonymous role used here because it's silly to name such a small role.
Radvendii: helps to know that roles are used for runtime overriding of methods. 20:01
Radvendii oh... i guess that does make sense
TimToady
.oO( All of life is a stage, and we all play many anonymous roles... )
20:01 Vlavv` joined
Radvendii TimToady: :P 20:01
20:01 baest left
moritz nom: class A { method x() { }; method x() { } } 20:02
p6eval nom d83a1c: OUTPUT«===SORRY!===␤Package 'A' already has a method 'x'; did you mean to declare a multi-method?␤»
moritz nom: sub a { }; sub a { }; 20:03
p6eval nom d83a1c: OUTPUT«===SORRY!===␤Illegal redeclaration of routine 'a' at line 1, near ";"␤»
TimToady I usually put questions like that in parens
so that the ? doesn't look funny with " at foo line 42" after it
masak submits rakudobug
jnthn heh 20:04
TimToady realizes he never made that nano-principle clear... :)
jnthn nom: say [^^] 5, 9, 17
p6eval nom d83a1c: OUTPUT«17␤»
jnthn [^^] is a curious beast.
masak TimToady: there are many good things to steal from Perl 5's error messages.
TimToady nom: say 5 ^^ 9 ^^ 17 20:05
p6eval nom d83a1c: OUTPUT«Nil␤»
masak I prefer Nil to 17.
jnthn 5 ^^ 9 gives False. False ^^ 17 of course gives 17...
Radvendii what is ^^?
masak Radvendii: eyebrows.
:P
Radvendii haha
TimToady short-circuit XOR
but it's a list infix
jnthn masak: Well, same. Any idea how we get reduce to do the right thing? :) 20:06
TimToady so it's only true if there's one true value
nom: say 5 ^^ 9 ^^ 17 ^^ 'phooey'
p6eval nom d83a1c: OUTPUT«Nil␤»
TimToady nom: say [^^] 5, 9, 17, 'phooey'
p6eval nom d83a1c: OUTPUT«Nil␤»
TimToady nom: say [^^] 5, 9, 'phooey'
p6eval nom d83a1c: OUTPUT«phooey␤»
Radvendii haha, weird 20:07
jnthn TimToady: Only happens with odd number of 'em. :)
masak jnthn: none whatsoever. ISTR hearing about the problem years ago.
jnthn: like, 2009.
jnthn TimToady: You can probably imagine why it's getting the result it is... :)
20:07 baest joined
TimToady so it's right half the time, which is better than a stopped clock, barely 20:07
masak "Perl 6: better than a stopped clock, barely" 20:08
TimToady I can imagine, sure, but [op] 1,2,3 is defined to be the same as 1 op 2 op 3
moritz modulo short-circuitry
colomon niecza: say [^^] 5, 9, 17
jnthn Right.
p6eval niecza v13-332-g677cf15: OUTPUT«Unhandled exception: System.Exception: Unable to find lexical &infix:<^^> in mainline␤ at Niecza.CLRBackend.NamProcessor.ResolveLex (System.String name, Boolean upf, System.Int32& uplevel, Boolean core) [0x00000] in <filename unknown>:0 ␤ at Niecza.CLRB…
TimToady niecza: say 5 ^^ 0 20:09
p6eval niecza v13-332-g677cf15: OUTPUT«Unhandled exception: System.Exception: Unable to find lexical &infix:<^^> in mainline␤ at Niecza.CLRBackend.NamProcessor.ResolveLex (System.String name, Boolean upf, System.Int32& uplevel, Boolean core) [0x00000] in <filename unknown>:0 ␤ at Niecza.CLRB…
TimToady hah
jnthn The spec says that ^^ doesn't short-circuit.
er
The spec says that [^^] doesn't short-circuit.
TimToady doesn't thunk
Radvendii nom: @a does role { at_pos
p6eval nom d83a1c: OUTPUT«===SORRY!===␤Variable @a is not predeclared at line 1, near " does role"␤»
Radvendii aaah!
premature send
moritz but still the error was right :-) 20:10
TimToady jnthn: the spec doesn't say it doesn't short-circuit; it says it doesn't thunk its args, S03:1377 20:11
Radvendii nom: my @a does role { at_pos($j) { say "test" } }; @a[0];
p6eval nom d83a1c: OUTPUT«===SORRY!===␤Variable $j is not predeclared at line 1, near ") { say \"t"␤»
Radvendii yeah, that's the error i'm getting
jnthn TimToady: ah, I see
Radvendii but $j is an argument...
colomon is trying to remember the details of niecza doing something funky with reduce and list ops.... 20:12
TimToady perl6: say [||] 1, 0 xx *; # should terminate
benabik Radvendii: try method at_post 20:13
*at_pos
p6eval niecza v13-332-g677cf15: OUTPUT«===SORRY!===␤␤This macro cannot be used as a function at /tmp/_KOy8dLfF6 line 1:␤------> say [||] 1, 0 xx *⏏; # should terminate␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line …
..pugs b927740, rakudo d83a1c: OUTPUT«(timeout)»
TimToady perl6: say (1, 0 xx *)[0]; # should terminate
p6eval pugs b927740, rakudo d83a1c, niecza v13-332-g677cf15: OUTPUT«1␤» 20:14
Radvendii benabik: that works
TimToady reductions on list associative operators must take that into account 20:15
masak perl6: multi postfix:<!>(0) { 1 }; multi postfix:<!>(Int $n) { $n * ($n - 1) }; say 5! 20:21
p6eval rakudo d83a1c, niecza v13-332-g677cf15: OUTPUT«20␤» 20:22
..pugs b927740: OUTPUT«*** ␤ Unexpected "0"␤ expecting formal parameter or ")"␤ at /tmp/WnOGKkiQFs line 1, column 19␤»
masak augh
perl6: multi postfix:<!>(0) { 1 }; multi postfix:<!>(Int $n) { $n * ($n - 1)! }; say 5!
p6eval rakudo d83a1c, niecza v13-332-g677cf15: OUTPUT«120␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "0"␤ expecting formal parameter or ")"␤ at /tmp/N95bZ46NKD line 1, column 19␤»
masak niecza: multi postfix:<!>(0) { 1 }; multi postfix:<!>(Int $n) is looser(&infix:<+>) { $n * $n - 1! }; say 5! 20:23
p6eval niecza v13-332-g677cf15: OUTPUT«(timeout)»
masak hm.
niecza: multi postfix:<!>(0) { 1 }; multi postfix:<!>(Int $n) is looser(&infix:<+>) { $n * ($n - 1!) }; say 5!
p6eval niecza v13-332-g677cf15: OUTPUT«(timeout)»
masak hm. 20:24
moritz maybe that tries to install two postfixes, one tighter than +, one looser
Radvendii hm... is there any way to give something access to private variables without making them public?
something specific, that is
masak I think multi subs shouldn't accept looser/tighter/equal at all. only only and proto should. 20:25
Radvendii: yes. a method.
moritz masak: aye
20:25 fridim_ joined
masak Radvendii: methods are the way to provide specific access to attributes. that's what mehtods do. 20:25
Radvendii masak: got it.
colomon niecza: multi postfix:<!>(0) { 1 }; multi postfix:<!>(Int $n) { $n * ($n - 1!) }; say 5!
p6eval niecza v13-332-g677cf15: OUTPUT«(timeout)» 20:26
dalek ecs: c6c7fcf | larry++ | S03-operators.pod:
clarify semantics of [op] for list associatives
benabik colomon: ($n - 1!) ?
colomon niecza: multi postfix:<!>(0) { 1 }; multi postfix:<!>(Int $n) { $n * ($n - 1)! }; say 5!
p6eval niecza v13-332-g677cf15: OUTPUT«120␤»
colomon benabik: just a cut-n-paste from masak, but yeah, that's his problem
Radvendii if i wanna just make them public, do i say "has $a=5 is rw;" or "has $a is rw = 5;"?
masak colomon: right, but mine was s'posed to be looser than infix:<-> 20:27
Radvendii hmm... it doesn't seem to like either one 20:29
can i initialize an instance variable and make it rw?
benabik has $.a ? 20:30
Radvendii that's what i meant
moritz Radvendii: has $.x is rw = 5;
benabik nom: class A { has $.a = 5 }; A.new.a = 4
p6eval nom d83a1c: OUTPUT«Cannot assign to a readonly variable or a value␤ in block <anon> at /tmp/DV29G8tr7m:1␤ in <anon> at /tmp/DV29G8tr7m:1␤»
benabik nom: class A { has $.a is rw = 5 }; A.new.a = 4 20:31
p6eval nom d83a1c: ( no output )
benabik nom: class A { has $.a is rw = 5 }; my $a = A.new; say $a.a; $a.a = 0; say $a.a
p6eval nom d83a1c: OUTPUT«5␤0␤»
benabik Radvendii: ^^ seems to work 20:32
Radvendii okay...
then my error must be coming from elsewhere
TimToady sure, ^^ works, but [^^] doesn't # rim shot
Radvendii TimToady: :P
it doesn't like it when i say method at_pos($s:, $i). what's wrong with that? 20:33
20:33 fridim_ left
Radvendii isn't that the syntax for getting self into $s? 20:33
masak no. 20:34
Radvendii :P
jnthn Don't need the ,
masak s/:,/:/
Radvendii got it
20:34 birdwind1pbird left
Radvendii that's odd... 20:34
why is that?
20:34 birdwindupbird left
arnsholt Well, the colon serves as separator, instead of the comma 20:36
Much like in normal text, really
Radvendii yeah, but programming isn't supposed to parse like english. i guess it doesnt really matter
benabik Radvendii: Perl _is_ supposed to parse like English. 20:37
moritz www.hirewolf.com/blog/hirewolf-open...ce-contest # might be interesting for some or the other p6 hacker :-) 20:38
Radvendii huh? since when. i thought that was python's gig. or applescript. perl has all of that punctuation.
moritz Radvendii: read "Programming Perl", section 27.2. "Perl Poetry" 20:40
TimToady why do you think punctuation is the opposite of English? Spoken English is full of punctuation...
moritz Radvendii: and also en.wikipedia.org/wiki/Black_Perl
TimToady wonders if ancient Hebrews actually thought their language didn't contain vowels... 20:42
20:42 mj41 joined
Radvendii that's hillarious (black perl) 20:43
TimToady: vowels were probably second-class letters. just like we dont think english contains a glottal stop (think of the - in uh-oh)
TimToady or the tt in glottal, if you're Cockney :) 20:44
Radvendii haha, yup. more people pronounce t in fountain that way
foun-n
TimToady funny that IPA seems to be a second-class citizen in Unicode 20:45
Radvendii how so? 20:46
if i "has" $.a and @.a, how does it know which "$obj.a" is refering to? 20:47
TimToady [fæunʔn] or some such, but that's latin 'ʔ'
moritz Radvendii: the answer is "don't do that"
Radvendii moritz: haha. okay
moritz: but what would it do? 20:48
moritz Radvendii: give them other names
dalek kudo/nom: 00e5547 | moritz++ | src/ (2 files):
throw X::Phaser::Once exceptions when two CATCH or CONTROL blocks appear in a block
kudo/nom: ef67fd4 | moritz++ | src/ (2 files):
more typed redeclaration exceptions
moritz you can always use singular and plural if you want
Radvendii nom: class A {has $.a=5; has @.a=(1,2)}; my A $a .= new; say $a.a;
p6eval nom d83a1c: OUTPUT«5␤»
Radvendii im just curious
interesting 20:49
TimToady e pluribus unum
dalek ast: 0b7d7d2 | moritz++ | S32-exceptions/misc.t:
more typed exception tests
Radvendii it's probably whichever you declare first
moritz $a.foo and $a.foos
Radvendii nom: class A {has $.a=5; has @.a=(1,2)}; my A $a .= new; say $a.a[0];
p6eval nom d83a1c: OUTPUT«5␤»
TimToady e pluribus unum -> $()
masak Radvendii: yes, but it's a bug.
Radvendii masak: oh, okay
moritz: i know, i'm just curious as to what it _would_ do
masak Radvendii: see rt.perl.org/rt3/Ticket/Display.html?id=108670 20:50
Radvendii alright, alright :P
benabik nom: class A {has @.a=(1,2); has $.a=5}; my A $a .= new; say $a.a;
p6eval nom d83a1c: OUTPUT«1 2␤»
benabik It is whichever was done first. 20:51
*declared
Radvendii benabik: yeah
why not just say $obj.$a
and $obj.@a
moritz eeks
because object names are identifiers
and if we allow sigils in identifiers, we lose the distiction between the two
Radvendii so?
moritz and then we don't know if $@a really means $( @a ), or a variable named $@a 20:52
20:52 fridim_ joined
moritz and ... a whole can of worms. 20:52
Radvendii it would only work directly after the . 20:53
$a.$b $a$b would still be illegal.
moritz and then somebody wants to override the accessors 20:54
jnthn hmmm...so: has $.x; has @.x; method x() { } # did we settle on this being allowed or disallowed?
jnthn can't quite tell from the ticket :)
moritz and wants to write method $x { } and method @x { }
also $a.$b is taken
Radvendii moritz: with what? 20:55
moritz nom: sub f($x) { say $x }; my $a = &f; 'foo'.%a
nom: sub f($x) { say $x }; my $a = &f; 'foo'.aa
nom: sub f($x) { say $x }; my $a = &f; 'foo'.$a
p6eval nom d83a1c: OUTPUT«===SORRY!===␤Confused at line 1, near "'foo'.%a"␤»
nom d83a1c: OUTPUT«Method 'aa' not found for invocant of class 'Str'␤ in block <anon> at /tmp/cxsR70bPFo:1␤ in <anon> at /tmp/cxsR70bPFo:1␤»
nom d83a1c: OUTPUT«foo␤»
Radvendii alright fine 20:56
be that way
;)
moritz is that way :-)
benabik nom: sub f($_) { .say }; 'foo'.&f # ? 20:57
p6eval nom d83a1c: OUTPUT«foo␤»
benabik \o/
Radvendii :P
brb guys 20:58
Radvendii is updating iTerm
20:59 Radvendii left, Radvendii joined
Radvendii o/ 20:59
TimToady jnthn: I think the consensus was leaning toward "allowed" 21:00
Radvendii TimToady: allowed but not for long i thoguh 21:01
moritz I think the consensus was that two attributes with accessors and the same short name were disallowed
because of .new(foo => $bar) ambiguty
Radvendii or "shouldnt be but is"
21:01 mj41 left
moritz has $.a; has @.a; ::?CLASS.new(a => 1) # which one is it? 21:02
TimToady not if there's a disambuating method explicitly put there
jnthn Locally:
> class A { has @.a = 2,3; has $.a = 1 }
Two or more attributes declared that both want an accessor method 'a'
Wordsmithing? :)
At the moment also:
> class A { has @.a = 2,3; has $.a = 1; method a() { } }
Two or more attributes declared that both want an accessor method 'a'
TimToady but I see the problem with BUILD args
jnthn Can easily change that.
But the .new/BUILD issue makes me prefer it as it is. 21:03
21:04 MayDaniel joined
moritz std: f($x!) { } 21:04
p6eval std 40d10f7: OUTPUT«===SORRY!===␤Variable $x is not predeclared at /tmp/S19nrfX0VF line 1:␤------> f(⏏$x!) { }␤Negation metaoperator not followed by valid infix at /tmp/S19nrfX0VF line 1:␤------> f($x!⏏) { }␤Parse failed␤FAILED 00:…
moritz std: sub f($x!) { }
p6eval std 40d10f7: OUTPUT«Potential difficulties:␤ $x is declared but not used at /tmp/icxc_zgq0E line 1:␤------> sub f(⏏$x!) { }␤ok 00:01 109m␤»
moritz std: sub f($x! = 3) { }
p6eval std 40d10f7: OUTPUT«===SORRY!===␤Cannot put a default on a required parameter at /tmp/ZwFtvK3hq3 line 1:␤------> sub f($x! = 3⏏) { }␤Potential difficulties:␤ $x is declared but not used at /tmp/ZwFtvK3hq3 line 1:␤------> sub f(⏏$x! = …
Radvendii why not? 21:05
moritz because you have to provide a value for it anyway
so what use is the default?
it just means you are confused
Radvendii oh
good call
TimToady we don't classify all unreachable code that way, but in this case it seems reasonable 21:06
jnthn: it seems to me that having the disambuating function able to "fix" it is necessary for composing two roles with conflicting attributes 21:08
masak that's essentially the same rationale for forbidding auto-generation of a public accessor that would never be generated.
moritz there's a huge number of errors related to signatures
because they are complex beasts, and much can go wrong
sleep& 21:09
jnthn o/, moritz
TimToady: Two roles with conflicting attributes will conflict on the attribute storage names even if accessor methods are resolved.
TimToady they're not stored with the sigil? 21:10
jnthn TimToady: oh, sorry
TimToady: OK, in that case they won't. :)
(Was thinking of the same sigil case...for some reason...)
21:12 mtk joined
dalek Heuristic branch merge: pushed 28 commits to nqp/qbootstrap by jnthn 21:12
21:14 fridim_ left 21:15 Radvendii left, MayDaniel left 21:19 molaf_ left 21:20 GlitchMr left
dalek kudo/nom: 71225d3 | jnthn++ | docs/ROADMAP:
A few difficulty estimates.
 21:21
kudo/nom: ccd75fa | jnthn++ | src/Perl6/Metamodel/AttributeContainer.pm:
Catch case of a class that does 'has $.x; has @.x'.
kudo/nom: 4ffbd61 | jnthn++ | src/Perl6/Metamodel/AttributeContainer.pm:
Allow an explicit method to resolve an attribute accessor conflict.
21:22 rindolf left 21:25 not_gerd left 21:29 Radvendii joined, kaare_ left
Radvendii im getting a weird maximum recursion levels error that is due to some behind the scenes method calling 21:30
where's that code-posting website thing?
colomon gist.github.com 21:31
21:33 bbkr1 joined
Radvendii gist.github.com/1665643 21:35
that's the code that's producing the problem
21:35 SHODAN left, BinGOs left, geekosaur left, SHODAN joined, BinGOs joined, geekosaur joined
Radvendii alright 21:36
i think i fixed that error...
but now I'm getting a weirder error.
website is updated... 21:37
it thinks i'm trying to access the method $s.a instead of the variable $s.a (@.a) 21:39
jnthn $s.h< $p > 21:42
That's probably not what you meant
$s.h{$p} # think you wanted this
the ... in $h<...> is always literal
Radvendii o-oh 21:46
i get the same error though... 21:47
(also, i thought <...> was double-quoted...)
the error is coming from the line "FETCH => method {$s.a[$i][$j]}," 21:48
is it possible that the $s is not following through to the "role{}" 21:49
jnthn Will that not mix the role in every time the variable is accessed? 21:50
Perhaps the issue is that you're expecting to get a fresh instance of the role each time you use it, though, and in reality roles are declarational so you're only getting one of them. 21:51
Radvendii mayeb... 21:52
but that still shouldn't be telling me that $s doesnt have a method "a"
right? 21:53
jnthn Note that $s.a[$i][$j] is a method call to the method a.
That's what accessors are - method calls.
Radvendii right. but it should automatically create that, no?
jnthn Radvendii: I suspect the issue isn't that it doesn't have one, it's that the $s inside the role doesn't work out. 21:54
Radvendii a-ah...
jnthn Radvendii: I suspect it's getting tied to the static lexpad or some such.
Radvendii so then how do i do this? making another class wouldn't work either... 21:55
would it?
jnthn Radvendii: No, but if you declare a my role MyAtPos[$s] { method at_pos ... } rather than the anonymous one 21:57
And then do ... does MyAtPos[$s]
Then you probably get what you're after.
Radvendii ah
okay
let me try that...
jnthn (that's a parametric role) 21:58
Radvendii where do i declare the role?
and what's parametric?
jnthn Radvendii: Maybe declare it just above your at_pos method (the outermost one) 21:59
Radvendii and what's that [$s] syntax...
jnthn Radvendii: Parametric means a role that takes some parameters
Radvendii ah
that answers my other question as well
jnthn :)
Radvendii if roles are very much like classes in the underlying code, can i make classes that take in parameters?
fsergot good night o/ 22:02
Radvendii o/
role test[$a,$b]{} 22:03
nom: role test[$a,$b]{}
p6eval nom 4ffbd6: ( no output )
jnthn Radvendii: They're not much like classes in the underlying code. :) 22:05
Radvendii oh. i thought i'd heard that 22:06
jnthn And no, you can't parameterize classes.
Radvendii dang :P
jnthn They have things in common, but plenty of things different too :)
Radvendii do i have to put a ';' at the end of a method declaration? 22:07
jnthn not if there's a new line after the } 22:08
masak wow, Regexp::Grammars has the new % syntax! 22:09
masak impressed
Radvendii jnthn: Weird.. this is probably just in the semantics, but when i try to access $a[0][1] (where $a is an object of the class), i get Proxy.new() 22:12
jnthn Radvendii: Hm. Sounds like somewhere along the way some kind of double-wrapping is going on. 22:14
Radvendii yeah, that's what i figured... 22:15
gist.github.com/1665643 is updated
at-pos is the same as at_pos right? 22:18
22:18 benabik left 22:19 benabik joined 22:21 alester left
Radvendii jnthn: is there a reason i can't manually call FETCH? 22:23
(on a proxy object)
22:23 cognominal_ left
Radvendii never mind 22:24
i can
jnthn Radvendii: I think your method at_pos's want to be "is rw" 22:25
22:25 cognominal joined
Radvendii i hate it when my methods make unions and have demands and all that :P 22:26
that doesn't seem to help...
bbkr1 I have difficult question that I need to solve for JSON-RPC refactoring. How to recognize dispatch failure? For example my server was given handler class with method foo(Int,Str){…} while someone called server with foo("bar","baz"). Can I recognize that dispatch is not possible without calling foo? Maybe somehow through meta? 22:27
22:27 Trashlord left
masak 'night, #perl6 22:27
Radvendii masak: o/
bbkr1 good night
jnthn bbkr1: Yeah, because Rakudo's MAIN implementation does it 22:28
22:28 renormalist joined
jnthn bbkr1: The .candidates_matching method in any Routine (sub or method) will do it 22:29
bbkr1 digs into spec 22:30
22:35 dbr left
Radvendii jnthn: woah. 22:36
dalek p/bs: 1d0b7f2 | jnthn++ | src/6model/sixmodelobject.h:
Give STables a slot for marking them with an SC.
p/bs: 8071678 | jnthn++ | src/6model/knowhow_bootstrapper.c:
Mark STables created during bootstrap with the bootstrap SC also.
Radvendii jnthn: if i try to assign to $a[0], it doesnt even call STORE
it calls FETCH about 5 times, and then raises an error because i'm trying to assign to a non-container 22:37
is there a problem with having multiple arguments to roles? 22:38
like role MyAtPos[$s, $i]
because my syntax hylighter goes screwy when i do that
jnthn Should be fine.
22:38 fridim_ joined
Radvendii okay. it's probably just the highlighter 22:39
but still, why doesn't it call STORE?
jnthn Not sure - I thought you had that bit working.
(the one level deep bit)
Radvendii i thought so too
bbkr1 jnthn: I've found Routine class in source code and read methods. But how can I get Routine object from class? 22:40
jnthn bbkr1: Oh!
bbkr1: TheClass.^find_method('foo')
bbkr1: Can be on an object also 22:41
Radvendii Oh! it's because i was saying "say $a[0]=1;" 22:42
that's odd.. it should still be doing $a[0]=1 first, which would call STORE 22:43
bbkr1 rakudo: class Foo { method bar (Int, Str) {}}; my $method = Foo.^find_method("bar"); $method.candidates_matching.say # empty parcel. weird 22:45
p6eval rakudo 4ffbd6: OUTPUT«␤»
Radvendii i don't understand why it's calling FETCH over and over again 22:46
and why 6 times?
that's so arbitrary
bbkr1 rakudo: class Foo { method bar (Int, Str) {}}; my $method = Foo.^find_method("bar"); $method.candidates_matching(Foo, 1, "x").perl.say # yay, jnthn++, thanks! 22:48
p6eval rakudo 4ffbd6: OUTPUT«(method bar(Foo , Int , Str , Mu %_!) { ... },)␤»
jnthn bbkr1: Well, you didn't pass in any arguments :)
bbkr1: yes, like that
Radvendii: It's not particularly surprising FETCH would be called multiple times.
Radvendii why?
i'm just accessing it once
jnthn Radvendii: Proxy is a scalar container. Various bits of the runtime - as you pass the container around - will want to decontainerize it. 22:49
Radvendii weird. but OK
jnthn Radvendii: For example, whenever they bind it to the symbol "self", they do so.
Radvendii so the first layer of at_pos is working 22:50
jnthn OK, time for me to take some rest. 22:54
Radvendii haha alright
jnthn 'night o/
Radvendii goodnight o/
SO |O| O|Z 22:56
oops
did not mean to hit enter :P
23:04 Radvendii left 23:12 fridim_ left 23:18 whiteknight joined 23:23 spine left 23:29 crashtest is now known as autark 23:31 tokuhirom joined 23:34 autark is now known as jaschob 23:35 jaschob is now known as autark 23:37 tokuhirom left 23:38 tokuhirom joined 23:50 nebuchadnezzar left
sorear good * #perl6 23:50
TimToady: I do not understand very well how non-flattening is supposed to work 23:54
moritz: on a candidate set, niecza only respects precedence for the proto 23:55
23:56 nebuchadnezzar 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!