»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
Mouq m: say { :in{1+1}, :start{2+2} }.perl 00:09
camelia rakudo-moar c19ae7: OUTPUT«{"in" => -> ($_? is parcel) { #`(Block|139784554724576) ... }, "start" => -> ($_? is parcel) { #`(Block|139784554725048) ... }}␤»
Mouq m: say { :in{1+1} :start{2+2} }.perl
camelia rakudo-moar c19ae7: OUTPUT«{"in" => -> ($_? is parcel) { #`(Block|140100960787408) ... }}␤»
rindolf Hi all. 04:23
atroxaper_ Hello, #perl6 ! 07:58
yoleaux didn't pass me message from Mouq :( 07:59
.tell Mouq I merged your pull request and did some changes. Today I'm going to make a lists render. 08:00
yoleaux atroxaper_: I'll pass your message to Mouq.
dalek ast: 9618ad9 | (Elizabeth Mattijsen)++ | S02-types/deprecations.t:
Make %h = {...} deprecation test pass on parrot
 08:15
dalek kudo/nom: 4fd4639 | (Elizabeth Mattijsen)++ | src/core/Deprecations.pm:
Make deprecation message insensitive to hash order
 08:20
lizmat which should fix deprecation message tests on jvm 08:21
masak good antenoon, #perl6 09:01
atroxaper_ masak: hello! 09:03
timotimo ohai masak 09:18
dalek ast: b40d639 | (Elizabeth Mattijsen)++ | S22-package-format/local.t:
Argh, fudging cannot handle trailing ;
 09:38
ast: c4b7322 | (Elizabeth Mattijsen)++ | S22-package-format/local.t:
Apparently we can't do signals on JVM either
 09:41
masak dreams of a day when the fudger isn't text-based but based on the parse tree 10:11
dalek kudo/nom: c0eef70 | (Elizabeth Mattijsen)++ | src/core/CompUnit.pm:
Make method names a bit more consistent
 10:20
dalek kudo/nom: 0c38be3 | (Elizabeth Mattijsen)++ | src/core/CompUnit.pm:
Be at least internally consistent
 10:29
masak oh, and the whole fudger is ~250 lines of Perl 5 :) that's gotta change, too. 10:33
timotimo you reckon it'll only be 25 lines of perl6 when you're done refactoring? :) 10:34
masak well, challenge accepted.
probably more like 100 lines, though.
vendethiel- watches over masak
masak I wonder what an easy way would be to show it's in feature parity. 10:35
vendethiel- didn't quite get that last sentence
masak probably simply to register all the fudged files and compare them.
masak vendethiel-: if you replace a p5 fudger with a p6 fudger, you don't want them to differ in the output they produce. 10:36
lizmat s/feature parity/same features/
vendethiel- oh. I thought you mean p5 and p6 ¬¬ 10:36
vendethiel- was thinking of something akin to github.com/satyr/coco/wiki/additions
lizmat fwiw, I think there is something fundamentally wrong in the way that we create pre-compiled versions of modules 11:04
lizmat Instead of a pre-compiled version of Foo.pm called Foo.moarvm 11:04
masak ok?
lizmat it should be called Foo.pm.moarvm 11:05
this will preserve the original extension of the module in the filename
masak worksforme
that suggestion, I mean.
lizmat ok
I hope jnthn / moritz / tadzik / FROGGS will also agree 11:06
moritz +1 11:06
lizmat it should allow v5 and v6 modules with the same name (eg. Foo.pm and Foo.pm6) to co-exist in the same dir as pre-compiled files 11:07
dalek kudo/nom: 1154641 | (Elizabeth Mattijsen)++ | src/core/CompUnit.pm:
Make sure we have a Bool value
 11:14
dalek ast: ebc873e | (Elizabeth Mattijsen)++ | S22-package-format/local.t:
Some more CURLF / CompUnit tests
 11:30
jnthn I don't see people saying C compilers are fundementally wrong because they produce foo.o instead of foo.c.o :P 11:47
vendethiel- can you use v5; inside of a c compiler ? 11:49
.oO( This would explain so much, yet so little )
 11:50
lizmat jnthn: well, true but in C, you only have .c
in Perl 6, we already have .pm, .pm6
and .pl
vendethiel- (and .t, does those count ?)
lizmat no 11:51
because even if we would do a pre-compiled version of a test file
it would not work because most likely we would have a different version of the executor running against it 11:52
jnthn It'd be odd to have a .pm6 and a .pm both intended for Perl 6... :)
lizmat I mean, if neither the test-file or the executor changed
jnthn Yes, pre-compiling tests sounds kinda useless :) 11:53
lizmat $ ls lib/Test.pm
lib/Test.pm
jnthn Yes, and there's not a Test.pm6 there too :)
Anyway, I don't really object to it being Foo.pm.jar or Foo.pm.moarvm; if Rakudo's managing it then it can be called whatever we want it to anyway. 11:54
Though inferring the .pm file name from the compiled output name feels a tad fragile.
jnthn (If that was the plan, anyway.) 11:55
lizmat well, yes, that was the plan 11:57
feels less fragile then having to guess complety, no ?
*completely
jnthn feared so, and wonders why the more robust suggestion he made a few days ago has been ignored.
lizmat then that must have slipped my mind... :-(
jnthn Or does that not make the filename available early enough? 11:58
lizmat ah, you mean the $?FILES idea?
jnthn $?FILE, but yeah
lizmat well, two things:
jnthn The module loader gets back UNIT of the module
lizmat but that's just it: you would need to load the module 11:59
until you find out what the original filename was
s/until/before
jnthn Well, this is all very XY
I guess my real question is, why does the inference go in this direction...
I'd thought the .pm/.pm6 files were primary, and pre-compiled things serve as a kind of "cached" compilation of the module... 12:00
lizmat but you would want to attempt to load the cached version first, right?
jnthn So if you're loading a .moarvm file you already know it's the file for a particular comp unit.
lizmat but that's just it: you currently cannot be sure 12:01
if you have a Foo.pm, a Foo.pm6 in the same dir
one for v5, one for v6
then currently you can precomp only one of them 12:02
jnthn then you're probably crazy...but yes, you're right. :)
lizmat I'm just reminded of Murphy
anything that *can* go wrong, will
jnthn Well, DIHWITD :)
But yeah, no real objections to including the full module filename in the precomp name. 12:03
lizmat ok, then I'll try to get that working without breaking anything 12:07
BTW: $ ls lib/lib.pm6 12:10
lib/lib.pm6
so, 'use lib' refers to lib/lib.pm6, and 'use Test' refers to lib/Test.pm 12:11
jnthn aye :)
lizmat seems inconsistent to me... any reason for this ?
jnthn Well, I think we'll just have to live with people using both extensions.
lizmat agree, but do we need to be inconsistent in the core modules ?
jnthn Well, it lets us know pretty soon if we break one of the ways :D 12:12
lizmat that's true
:-)
jnthn I think the history of it is simply that Test.pm has been there longer than we have used .pm6 though. 12:12
jnthn So there's no deep reason or anything, afaik. Just history. 12:12
vendethiel- m: class A { method foo { say $.a } } 12:22
camelia ( no output )
vendethiel- m: class A { method foo { say $.a } }; A.new
camelia ( no output )
vendethiel- m: class A { method foo { say $.foo } }; A.new.foo 12:23
camelia rakudo-moar 0c38be: OUTPUT«(timeout)»
jnthn So recursion... 12:24
vendethiel- much deep. 12:25
lizmat on recursion, probably: 12:36
gist.github.com/lizmat/3c4246c57f0a695c7d21
a simple syntax error (forgot ;)
make settings compile go into oblivion 12:37
lizmat not sure if something should be done about it though, but could be helpful to know if your working on the settings 12:37
and all of a sudden, it doesn't complete compilation anymore 12:38
now first: cycling& 12:40
ChoHag Hah! The bug only shows up when the output is not tee'd into a log file. 13:04
dalek kudo-star-daily: e35ef30 | coke++ | log/ (13 files):
today (automated commit)
 15:20
rl6-roast-data: ec1e3b8 | coke++ | / (4 files):
today (automated commit)
[Coke] (fudger based on the parse tree. wouldn't that raise the bar too high for new implementations?) 15:21
psch wonders how useful loading classfiles from $CLASSPATH actually is for j-rakudo 16:09
i've hacked it into BootJavaInterop to load those, i.e. 'CLASSPATH=. perl6-j -e"use A:from<java> ..."' where A.class is in . works 16:10
previously it only worked with appending . to -Xbootclasspath, iirc
but then, things usually come jar'd anyway, don't they. i haven't actually tested how that works at the moment, but istr that vendethiel-++ had some fun with a friend with that 16:11
gist.github.com/peschwa/0158083bc4e5613a6390 # this is the patch, for reference
vendethiel- psch: yeah, though I think he stopped to basic stuff like io / stringbuffers, etc
(and it really wasn't pretty :P) 16:12
psch vendethiel-: did you get beyond loading anything that's not loaded by the java bootclassloader? i.e. stuff that's not in the jre 16:12
vendethiel psch: no clue, I'd need to ask him. But I don't think so.
I really just skimmed over what p6 brings to the table and we hacked for like 1h, so no dwelving deeper 16:13
vendethiel spent again 3 hours today explaining P6 with his game of life as an example
psch right, thanks for the info :)
the bigger blocker for neat jvminterop is definitely getting pretty signatures for resolving overloaded methods 16:14
in any case, with my patch one *could* write java classes that don't utilize overloading and use those from perl6
and it'd be all clean
but the use still seems limited to me, tbh :) 16:15
s/use/usefulness/
psch because java without any overloading that you probably wrote yourself could just be ported... :D 16:15
vendethiel psch: the usefulness of ? writing P6 class from jvm ?
well, but then, there's a reason we write p6 and not java code ;-) 16:16
he (my friend) was especially interested in grammars, that could make some of his code at least 10 times simpler ...
psch vendethiel: right, i better tosh the patch - it'd just encourage java people *not* to rewrite everything in perl6 ;) 16:17
vendethiel psch: why not ?
psch vendethiel: because it's a bit of an improvement for interop, i.e. using the existing java code 16:18
it was a bit of a silly joke
no one likes rewriting stuff just because
vendethiel I disagree, we do that all the time ;-)
psch oh, okay. shows what i know :) 16:19
vendethiel
.oO( I'm okay with them rewriting anything, so long they don't bring those stinky' factories along :P )
 16:21
psch vendethiel: bash.org/?946461 # you reminded me of this :) 16:31
vendethiel if only it were just a joke ... :( 16:32
psch i've only spend about a year actually working with java, in a rather small shop. over-abstraction didn't really happen there. 16:33
vendethiel well, factories were only created because of the lack of lambdas. 16:35
(java's finally catching up with this kind of stuff, after c++ even ...) 16:36
psch what i still want to get into jarakudo is patching -I and PERL6LIB into the classpath 16:40
but afaiu that means delegating *all* classloading to a custom classloader
as in, the patch above just enables CLASSPATH for perl6-j, but -I and PERL6LIB don't get searched for classfiles 16:41
psch which is obviously not intuitive at all. on the other hand it's also a bit arcane. 16:41
psch (i also have no idea in how far lizmat+ 16:42
(i also have no idea in how far lizmat++'s work on compunits might do things differently there)
(...but if i'm not completely dumb compunits don't play a role in loading :from<java> :) 16:44
lizmat is backlogging 16:48
grondilu off topic: antares launch in 3 minutes: spaceflightnow.com/antares/orb2/status.html 16:49
lizmat psch: actually, I'm thinking about creating an API in which class loading can be plugged 16:59
psch lizmat: 'class' here is the p6-construct? 17:01
psch i'm not sure i'm well-versed enough in the abstract concepts - i just stumbled upon something i thought should work and made it work as i thought it should 17:06
lizmat the problem with all of this, is that in the Perl world, loading a module is really loading a piece of code that hopefully contains that module 17:12
psch in any case i do recognize my patch as band-aid-y - as mentioned, using -I or PERL6LIB instead of CLASSPATH is probably neccessary
lizmat and that currently all of the module loading code lives in nqp or below that 17:13
lizmat I'm trying to get this to Perl 6 land as much as possible, so that a pluggable Perl6 API becomes more feasible 17:14
psch lizmat: how does this apply to loading java classfiles? should a perl6-m on encountering a use ..:from<java>; check for perl6-j? 17:16
lizmat a perl6-m should probably die on seeing a use ...:from<Java> 17:27
timotimo ideally, it'd work :P 17:29
psch in turn that means that a perl6-j encountering a use ...:from<java> checks if it has a .jar and .class capable loader plugged, which goes into Java land to load the class as java and then builds a FFI or somesuch onto the loaded class, right?
timotimo the FFI stuff already exists 17:31
timotimo it introspects the class you've pulled in and generates java code that knows how the call convention of rakudo/nqp works and calls the appropriate method on the passed instance 17:32
timotimo currently it is only able to figure out the right candidate to call on an "overrided" method if the method is the only one with the given number of arguments 17:34
at least last time i touched the code it was like that
you can find that stuff in the nqp repo somewhere under vm/jvm/.../runtime/...very_long_path.../
psch timotimo: right, i did see the existing FFI stuff, or rather played with it a bit 17:37
that's where i actually came across the need of not wanting to add '.' to the -Xbootclasspath option for perl6-j
lizmat is a complete jvm noob 17:41
timotimo lizmat: i was a complete jvm-internals noob before i did that :) 17:43
psch i think that's a distinction to make here though: does pluggable class loading handle loading native code for the corresponding backend? i.e. the java example i already gave, potential PIR code on parrot and non-perl6 moar bytecode? 17:51
psch i don't know if that's been thought of or really seems reasonable, but i think it's worth mentioning... :) 17:53
i do actually agree, to an extend, with timotimo's "ideally it would work" with the added caveat "provided perl6-j is available"
timotimo oh
why have a perl6-j, why not just bundle our own sleeker jvm, like Avian?
:P
psch i think you just dropped that ball in your court :) 17:54
if only because i'm also quite a jvm noob 17:55
dalek kudo/nom: b25b868 | (Tobias Leich)++ | src/Perl6/World.nqp:
add nqplist helper hllsym for use in slangs
 18:51
Ven m: role WithValue { method !m { say $.a } }; class B does WithValue { has $.a = 5; method f { self!m } }; say B.new.f; 19:02
camelia rakudo-moar 115464: OUTPUT«5␤True␤»
Ven so, a class can access the private methods of a role 19:03
m: role WithValue { has $!a = 5; }; class B does WithValue { method f { $!a } }; say B.new.f;
camelia rakudo-moar 115464: OUTPUT«5␤»
Ven and the fields too. But the inverse is untrue (a role can't accede private stuff from a class) 19:03
Ven Well, it's really hard to explain textually. I'm afraid this learnXinY will actually contain so much, anybody trying to understand it will have to re-read it a few times ... 19:09
Well, I guess I'll go with that for now, and change it to be slower-paced down the road, when it's more advanced
cognominal FROGGS, I am very excited about your patches about slangs but I don't see the general picutres to make use of slangs. 19:18
masak Ven: one way to view that is, by the time the class is declared, it "knows" about the role and all its components. but the reverse isn't true -- a role, as it is being defined, doesn't "know" about the class(es) it gets mixed into, and so cannot refer to their private things.
Ven: that reasoning works, though it's slightly specious because it reminds me of class inheritance.
Ven masak: don't get me wrong -- I understand *why* (and am grateful for that ;).), it just makes a tiny bit harder still to explain 19:19
When I try and read stuff, I just feel like there's so much crammed in it ..
masak a rationale/explanation was what I was aiming to provide above. 19:20
FROGGS cognominal: DSL is all I'm saying :o) 19:22
Ven masak: yeah, I have something like that too, but then again ... It feels like a big chunk of explanations
p6 is really complex a language.
cognominal FROGGS: I suppose there is code that motivates these patches. 19:23
FROGGS cognominal: v5, yes
cognominal ho, thx.
FROGGS++ 19:24
awwaiid Ven: one thing that I've found is that p6 tends to be a superset of other things. So if you just focus on, say, the bits that are similar to Ruby, then you can progress pretty quickly. 19:25
It is just easy to get distracted by the SuperStuff 19:26
Ven yeah, but I can't assume knowledge from other languages.
For me, I did indeed relate to previous knowledge, but I can't "project" that onto my readers 19:27
awwaiid Maybe you can assume the same progression as other languages. So like stealing the structure of a tutorial in one of the other dynamic languages
awwaiid but I missed it -- what are you up to? 19:28
Ven awwaiid: a perl 6 "learnxinyminutes" 19:42
Ven I have the operators down, the semantics basics, and I'm making good progress on classes. 19:43
It's just that p6 is complex (to be easy)
m: sub foo(Int $ where * > 10) { True }; sub a { foo(5) }; 19:56
camelia ( no output )
Ven checks if infix:<*> is pure 19:57
no, it's not. Any reason it's not ? I have a feeling we could check a lot more stuff at compile-time with pure markers
(ie, if we know the where only has pure calls, we can actually just check at compile time) 19:58
pmurias is 'is pure' specced? 20:03
Ven is there a difference between anon sub {} and -> {} ? 20:11
jnthn return 20:12
moritz sub returns a Routine, which is a return() boundary 20:15
awwaiid innnteresting 20:16
hm! that's very intersting. and subtle 20:18
Mouq which means that you can have closures from which you can return out of the outer function
yoleaux 08:00Z <atroxaper_> Mouq: I merged your pull request and did some changes. Today I'm going to make a lists render.
timotimo it's a very useful distinction, and easy to remember if you think of sub filter(@bar) { for @bar -> $foo { return $foo if m/hello/ } }
timotimo in this case you don't want the return inside the for loop to simply skip to the next iteration - in this case you want "next" instead 20:19
Ven uh, that's the diff between -> {} and {}, isn't it ?
timotimo huh?
Ven m: sub t(@a){map(-> { return True if $_ }, @a)}; t((1, 2, 3)); 20:20
camelia rakudo-moar b25b86: OUTPUT«Unhandled exception: Too many positional parameters passed; got 1 but expected 0␤ at /tmp/uD_2gT1o6M:1 (<ephemeral file>::0)␤ from src/gen/m-CORE.setting:7736 (/home/p6eval/rakudo-inst-2/languages/perl6/runtime/CORE.setting.moarvm::631)␤ from src…»
timotimo m: say &map.signature.perl 20:20
camelia rakudo-moar b25b86: OUTPUT«:(Any |)␤»
timotimo heh. helpful
Ven ._.
awwaiid m: sub f { my $x = -> { return 5; } ; $x() ; 10 } ;sub g { my $x = sub { return 5; } ; $x() ; 10 } ; say "f=&f() g=&g()"
camelia rakudo-moar b25b86: OUTPUT«f=5 g=10␤»
Ven this segfaults on my REPL .... 20:21
so, what's the difference between -> {} and {} then ? "just" for arguments ?
I guess I got it confused with ruby's distinction between {} and ->{}
Mouq doesn't understand the ruby closure syntax, but he hasn't looked much into it 20:23
Ven nobody understands it, we're just faking along at that point.
Plus, it changes between 1.9 and 2.0, iirc
timotimo -> ... { } and { } are the same, except the second may be interpreted as a hash if it has the "right" (or "wrong") syntax/content
Ven updates his tutorial accordingly 20:24
Ven added a bunch of stuff 20:28
multi methods, "done", classes, kinda done. 20:29
vendethiel (thanks for your quick answer ! It'd have been pretty bad to fix later on) 20:33
timotimo oh no 20:34
thank you
tutorial work is quite valuable in perl6's current economy
Ven (proof-reading *very* welcome, at least on the structure for nwo) 20:37
masak 'night, #perl6 21:00
dalek ast: 91cf966 | coke++ | S02-names-vars/variables-and-packages.t:
Fix scoping on sub.

  jnthn++		 23:05
rl6-bench: 20f39e2 | (Geoffrey Broadwell)++ | analyze:
Stub in summarize_results_html_history
 23:08
rl6-bench: 084db6c | (Geoffrey Broadwell)++ | microbenchmarks.pl:
Add 'bind' tag to two microbenchmarks
rl6-bench: 0e5a1cf | (Geoffrey Broadwell)++ | / (3 files):
Two refactors: move summary common code; stop overusing 'perl'

Move much of the common setup done by all (non-JSON) summarizers out to a separate module, and just use it from the various summary routines.
Also stop overusing 'perl' where I really mean 'compiler'. Not only is it confusing now, it will get much worse if I ever add non-Perl-family languages to the compiler suite.
rl6-bench: 0663943 | (Geoffrey Broadwell)++ | components.json:
Merge branch 'master' of github.com/japhb/perl6-bench
colomon_ japhb++ 23:20
yoleaux 11 Jul 2014 20:24Z <[Coke]> colomon_: github.com/sorear/niecza/issues/191
dalek ast: 88239a0 | coke++ | S06-routine-modifiers/lvalue-subroutines.t:
unfudge passing tests
 23:49
[Coke] we are not running any of the S01-perl5-integration tests, but there are fudge directives in them: thoughts on keeping versus removing the fudge? 23:54
I will just ignore them for now. 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!