»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 moritz on 3 May 2013.
colomon .ask masak github.com/masak/html-template/pull/6 ..... 01:26
yoleaux colomon: I'll pass your message to masak.
colomon resolve stage failed for Rakudo::Debugger: Could not find file 'Build' for module Build 02:32
muixirt good morning 06:00
sorear o/
muixirt why isn't the first exmple in classtut working anymore? 06:01
Teratogen tut tut 06:02
sorear what are you doing here 06:04
Teratogen hanging out
how are you doing this evening, sorear? 06:05
=)
muixirt (doc.perl6.org/language/classtut)
sorear good 06:06
moritz muixirt: because it uses the default constructor to initialize private attributes; see the discussion yesterday 07:07
the docs need to be fixed
dalek rl6-bench: bc2907d | (Geoffrey Broadwell)++ | / (2 files):
Support setting --enough-time for scalable tests
 07:33
rl6-bench: 84f9723 | (Geoffrey Broadwell)++ | / (4 files):
Add ability to ignore compile time, by doing '0 scale' runs before main scaled timing loop, and then later subtracting that time back out; refactor scaled timing loop to be more flexible and clearer; fix a microbenchmark that would loop forever with 0 scale
FROGGS ohh, fuzzy logic 07:36
dalek p: c557c27 | moritz++ | Configure.pl:
be smarter about guessing Darwin architecture

Closes #100		 07:50
dalek rl6-bench: 946f2d0 | (Geoffrey Broadwell)++ | / (2 files):
Support axis labels for plots, including custom labels for rc-forest-fire and rc-man-or-boy-test
 08:05
muixirt sorry moritz , was afk 08:17
labster good morning 08:18
moritz muixirt: no problem 08:19
good morning labster, *
muixirt moritz: the error message i got was: 'Nominal type check failed for parameter '@dependencies'; expected Positional but got Array instead' 08:20
and that seems not to depend on how 'dependencies' and 'callback' are declared 08:25
labster rn: my class A { has $.foo = "hi"; method foo { say "foo"; }; }; A.new.foo 08:28
camelia rakudo 25a0b3: OUTPUT«foo␤»
..niecza v24-51-g009f999: OUTPUT«===SORRY!===␤␤Two definitions of method foo (see line 1) at /tmp/9w2NDFWoPc line 1:␤------> class A { has $.foo = "hi"; method foo ⏏{ say "foo"; }; }; A.new.foo␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/ni…
labster although, that bug/feature is the only thing making my code work right now. 08:29
moritz muixirt: could you please open a ticket at github.com/perl6/doc/issues/new
labster: rakudo is correct there
labster cool, wasn't sure, didn't feel like looking it up :) 08:30
labster change one character in IO::Handle: $.path -> $!path ===> Cannot unbox a type object as a native str 08:42
moritz it probably means that $!path isn't initialized 08:44
and then it's Any
that's a type object
dagurval morning o/
moritz and some object tries to uses it as a string, unboxes, and its nose hits the road
\o morning dagurval
labster oh, that's fair, and $.path is a method, which isn't an Any... yet.
morning dagurval, working on your IO::Handle.path request now :) 08:45
dagurval labster: sweet! :)
moritz if you say 'has $.path', then it marks the attribute as public, meaning that a .new(path => $foo) can initialize it
so it doesn't just add an accessor, it also changes initialization 08:46
so even if you write a 'method path', it can make sense to declare the attribute as public
labster It also changes output: IO::Handle.new(ins => 0, chomp => Bool::True, path => IO::Path.new(basename => "INSTALL.txt", directory => ".", volume => ""))
moritz yes, .perl also hides private attributes by default 08:47
labster I'm wondering if it makes sense to change the name of the attribute. After all, a pipe isn't exactly a path. 08:48
I just can't think of a good name for it 08:49
moritz 'resource' # as generic as it goes
file-system-representation 08:50
moritz likes none of his ideas
labster Already came up with resource, also didn't like it.
labster Although, pretty much no one uses IO::Handle.new directly, to it might not matter. 08:51
*so it might 08:52
FROGGS there will be ppl who uses it directly, no doubt
dagurval people always use an API in ways you won't expect :)
labster No one complained when when we renamed it from IO to IO::Handle, and that was last month :) 08:52
Of course, we haven't had a star release in a while. 08:53
dalek rl6-bench: 2b06b5b | (Geoffrey Broadwell)++ | analyze:
Note in analyze output when compile time is being ignored
 09:00
rl6-bench: e00a844 | (Geoffrey Broadwell)++ | mi (2 files):
Set larger minimum scales to avoid noise at left of plots when compile time is ignored, potentially preventing nonsense performance ratios
muixirt rn: sub test($a, Int *@b) { say $a, @b; }; test(1,2) 09:24
camelia niecza v24-51-g009f999: OUTPUT«12␤»
..rakudo 25a0b3: OUTPUT«Nominal type check failed for parameter '@b'; expected Positional but got Array instead␤ in sub test at /tmp/iGxn8YtEWs:1␤ in block at /tmp/iGxn8YtEWs:1␤␤»
colomon rn: sub test(Int *@b) { say @b }; test(1, 2) 10:05
camelia niecza v24-51-g009f999: OUTPUT«1 2␤»
..rakudo 25a0b3: OUTPUT«Nominal type check failed for parameter '@b'; expected Positional but got Array instead␤ in sub test at /tmp/dSd_6C2Lnx:1␤ in block at /tmp/dSd_6C2Lnx:1␤␤»
colomon rn: sub test(*@b) { say @b }; test(1, 2)
camelia rakudo 25a0b3, niecza v24-51-g009f999: OUTPUT«1 2␤»
colomon known rakudobug? 10:06
timotimo is there a way, with macros perhaps, to have a call like "psay $0, $0.WHAT, $0[0].WHAT" display something like '$0: ' ~ $0.gist ~ ', $0.WHAT: ' ~ $0.WHAT.gist ~ ', $0[0].WHAT: ' ~ $0[0].WHAT; 10:30
? 10:31
muixirt colomon: well, yes: rt.perl.org/rt3//Public/Bug/Displa...?id=113964 10:37
but is it really only a LTA bug? 10:38
timotimo oh my. i could open the source file, figure out the $?LINE of the caller and read that out. 10:39
colomon muixirt: looks like a real bug to me, not a LTA error message bug. 10:40
especially because the LTA error message sounds like something very wrong is going on internally...
timotimo r: sub yoink($foo, *@bar) { say $bar.WHAT }; yoink("foo", "bar", "baz"); yoink("foo") 10:49
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Variable '$bar' is not declared␤at /tmp/VDAkzSJIaE:1␤------> sub yoink($foo, *@bar) { say $bar.WHAT⏏ }; yoink("foo", "bar", "baz"); yoink("f␤ expecting any of:␤ method arguments␤ postfix␤»…
timotimo r: sub yoink($foo, *@bar) { say @bar.WHAT }; yoink("foo", "bar", "baz"); yoink("foo")
camelia rakudo 25a0b3: OUTPUT«(Array)␤(Array)␤»
timotimo r: sub yoink($foo, *@bar) { say @bar ~~ Positional }; yoink("foo", "bar", "baz"); yoink("foo")
camelia rakudo 25a0b3: OUTPUT«True␤True␤»
timotimo huh ... !
timotimo r: class Task { method new($foo, Task *@d) { say @d.WHAT; say @d ~~ Positional; self.bless(*) } }; my Task $a = Task.new("a", Task.new("b"), Task.new("c")); 10:51
camelia rakudo 25a0b3: OUTPUT«Nominal type check failed for parameter '@d'; expected Positional but got Array instead␤ in method new at /tmp/lVB4e2lcSi:1␤ in block at /tmp/lVB4e2lcSi:1␤␤»
timotimo r: class Task { method new($foo, *@d) { say @d.WHAT; say @d ~~ Positional; self.bless(*) } }; my Task $a = Task.new("a", Task.new("b"), Task.new("c"));
camelia rakudo 25a0b3: OUTPUT«(Array)␤True␤(Array)␤True␤(Array)␤True␤»
timotimo rakudobug?
oh, that's even what you were discussing right before i looked at it 10:53
timotimo procrastinates from writing talk slides by perhaps adding an error to slurpy parameters 10:54
std: sub frob(Int *@foo) { };
camelia std 0336087: OUTPUT«Potential difficulties:␤ @foo is declared but not used at /tmp/GVO8V4tTuH line 1:␤------> sub frob(Int *⏏@foo) { };␤ok 00:00 44m␤»
timotimo std: sub frob(Int *@foo) { }; frob(1, 2, 3) 10:55
camelia std 0336087: OUTPUT«Potential difficulties:␤ @foo is declared but not used at /tmp/UzyLjzCOqB line 1:␤------> sub frob(Int *⏏@foo) { }; frob(1, 2, 3)␤ok 00:00 46m␤»
timotimo a spec change can come when my patch is complete, i guess.
or is this just an NYI warning that should be displayed?
moritz it's not at all clear how that one should work 10:56
I mean if you write sub f(Int @f) { }, then it only allows an Array[Int], not an Array containing only ints
now how do you check that for a slurply, lazily? 10:57
timotimo ah, that's a good point :( 11:00
i'll just .sorry('Typed slurpies will never work') for now, the exact wording and exception type and specciness is still up to discussion after the code is in. 11:01
timotimo builds and spectests 11:02
timotimo huh, there already is code for that 11:10
ah, yeah, because i dropped it in there (but it's wrong) 11:11
i should be making slides instead, i'm not finding the obviously correct position for the code yet 11:13
oh, apparently it's easier than i thought, and i looked at all the wrong places. silly me. 11:21
FROGGS timotimo: I know exactly what you mean :o) 11:22
bbkr moritz++ # darwin architecture detect (I will test it later this evening) 11:24
FROGGS bbkr: if it doesnt work for you then please paste the output of `ld`and `uname -a` 11:25
timotimo are the comments for karma modifications collected?
FROGGS timotimo: I believe karma counting doesnt work at all atm
karma timotimo
.karma timotimo
lizmat .karma kameleon
FROGGS *g* 11:26
lizmat++
timotimo .karma timotimo
can't tell if not implemented or implemented in perl6 %) 11:27
FROGGS I believe phenny was the one who could tell us 11:27
lizmat yoleaux: where's phenny ? 11:28
timotimo phenny only lived once i'm afraid :(
moritz if somebody wants to write some code that analyzes the IRC logs for karma, please do 11:29
lizmat r: class A {}; grammar A{} # does this also bomb with 2 UNIT's ?
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Redeclaration of symbol A␤at /tmp/qCiNcOmLVn:1␤------> class A {}; grammar A⏏{} # does this also bomb with 2 UNIT's ␤ expecting any of:␤ statement list␤ horizontal whitespace␤ postfix␤ …
FROGGS ahh, the clogs tell me that ilbot logs them, but it cant talk about it 11:30
moritz I can either run it on my machine, or provide sql dumps 11:30
FROGGS moritz: so my statement isnt true?
moritz FROGGS: which statement?
lizmat feels a philosophical debate forming 11:31
FROGGS <FROGGS> ahh, the clogs tell me that ilbot logs them, but it cant talk about it
moritz it's true
FROGGS but it doesnt count them? 11:31
moritz 13:29 < moritz> if somebody wants to write some code that analyzes the IRC logs for karma, please do
no. ilbot just logs, and does nothing else
FROGGS k 11:32
moritz but somebody[tm] could analyse the old logs to initialize some karma tracking bot with it
moritz (or update them from a nightly cron job, or so) 11:33
lizmat answer to my question: yes, but at run time: Merging GLOBAL symbols failed: duplicate definition of symbol A 11:34
.tell jnthn that there is already a fix for .exists on typed hashes: it's called github.com/rakudo/rakudo/pull/147 11:35
yoleaux lizmat: I'll pass your message to jnthn.
timotimo gist.github.com/timo/88c92ee2dcc5040a4f93 - spectesting this now; comments on typed exception or wording? 11:36
FROGGS timotimo: "... won't work" - why?, maybe write it as "... not supported." 11:37
lizmat one of my initial experiences with Perl 6 in the past months, was trying to grok Any and Mu
Mu to me means, something unknown 11:38
Any to me means, I don't care
so I started using Mu for some applications
timotimo spectests pass with the change, now to try some slurpies in different locations :)
lizmat whereas "normal" perl 6 users would probably just have used "Any" and be done with it
Juerd What's the semantic difference, lizmat? 11:39
lizmat If I want to store in a hash that something is unknown, I would use Mu
timotimo aha. doesn't alert with slurpy blocks
lizmat %version{$module}= Mu
timotimo er, of course it won't if i leave out the type info ...
Juerd lizmat: Not *? 11:40
lizmat if I want to load a module, and I don't care about its version, I'd specify Any
Juerd Never mind that, I forgot that * is a closure thing now.
timotimo wouldn't whatever be better?
lizmat well, Whatever also means some kine of Any to me
timotimo too late m)
lizmat and there's that :-)
Juerd whatever, mu, any... Do we really need this? :(
+cool
timotimo to me, there's the additional difference that Any is "an undefined value of any type" and whatever is more like "whatever value you come up with" 11:41
lizmat Ah, yes, well, I'm not qualified yet to give that answer :-)
Juerd This is one of the Very Hard parts of the language. Very abstract and hard to grok.
lizmat anyway, from a readability point of view, I decided to introduce "Unk"
constant Unk := Any;
whenever I need to indicate something unknown, I will be using Unk 11:42
Juerd Why not Unknown?
lizmat Unk is just another three letter word like Any
Juerd It's not a word
It's the result of truncating a word.
lizmat so you could argue that for Any(thing) 11:43
Juerd It moves the effort from writing to reading, while reading is done much more often.
I'm not going to discuss whether "any" is a word. It just is.
lizmat Int ? Str ? Bool ?
Juerd int, str and bool are well known abbreviations. 11:44
lizmat well, unk is as well apparently: acronyms.thefreedictionary.com/UNK 11:45
Juerd I'd love Integer, String and Boolean, but I can accept the shorter ones because they're decades old.
Far fetched.
lizmat anyways, this is not about bikeshedding: this was more about the idea of using another name for something we don't know in Perl 6 apart from Any 11:46
or Whatever *
timotimo i don't know about constant Unk := Any, because undefined Any is often used as a failure value 11:47
lizmat I thought Nil was ?
Juerd Any, Nil, Whatever, Mu, Cool, Unk. How many more? 11:48
I'm surprised
FROGGS r: say (1, Nil, 3).elems; say (1, Any, 3).elems; say (1, Mu, 3).elems; say (1, Cool, 3).elems; say (1, Whatever, 3).elems 11:50
camelia rakudo 25a0b3: OUTPUT«2␤3␤3␤3␤3␤»
moritz fwiw it might make sense to use something as a version marker that you can smart-match against
FROGGS all behave a bit different depending on its content
would be cool to have a matrix about that
moritz FROGGS: that particular behavior of Nil is going away
FROGGS ohh, good lord
moritz r: say 2 ~~ * 11:51
lizmat moritz: you mean matcing v1.0.1 ?
camelia rakudo 25a0b3: OUTPUT«True␤»
lizmat r: say 2 ~~ Any
camelia rakudo 25a0b3: OUTPUT«True␤»
Juerd FROGGS: Whenever a programmer needs a matrix to determine semantics, the design is flawed.
FROGGS true 11:51
lizmat r: say v1.0.1 > v.1.0.0 11:52
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Confused␤at /tmp/F_U1kWSfxb:1␤------> say v1.0.1 > v.⏏1.0.0␤ expecting any of:␤ dotty method or postfix␤»
lizmat r: say v1.0.1 > v1.0.0
moritz lizmat: true, the smart-match against Any is also true, but it's a type check
camelia rakudo 25a0b3: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: Mu *%_)␤ in method Real at src/gen/CORE.setting:870␤ in sub infix:<>> at src/gen/CORE.setting:3047␤ in sub infix:<>> at src/gen/CORE.setting:3045␤ in block at /tmp/hExiRcuKXA:1␤␤»
moritz lizmat: do you want a type check for version checking? 11:53
r: say v1.0.1 cmp v1.0.0
camelia rakudo 25a0b3: OUTPUT«Decrease␤»
lizmat I want a smartmatch on version numbers, so I can easily grep on it
moritz r: say v1.0.1 ~~ v1 11:54
camelia rakudo 25a0b3: OUTPUT«True␤»
moritz r: say v1.0.1 ~~ v1.0
camelia rakudo 25a0b3: OUTPUT«True␤»
moritz r: say v1.0.1 ~~ v1.0.0
camelia rakudo 25a0b3: OUTPUT«False␤»
moritz there you go
lizmat so you're saying that Already Works ?
then my work is done :-)
moritz yes, Version has smart-matching implemented
r: say v1.0.3 ~~ v2.*.3 11:55
camelia rakudo 25a0b3: OUTPUT«False␤»
moritz r: say v2.*.3.perl
camelia rakudo 25a0b3: OUTPUT«Version.new('2.*.3')␤»
moritz r: say v1.0.3 ~~ v1.*.3
camelia rakudo 25a0b3: OUTPUT«True␤»
lizmat cool 11:56
r: say v.7.1.2 ~~ v.6.2.3+ 11:57
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Confused␤at /tmp/JxSQnJZiV4:1␤------> say v.⏏7.1.2 ~~ v.6.2.3+␤ expecting any of:␤ dotty method or postfix␤»
lizmat S11:558 apparently NYI :-(
r: say v.7.1.2 ~~ v.6.2.3 .. *
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Confused␤at /tmp/ufPyAgWYeL:1␤------> say v.⏏7.1.2 ~~ v.6.2.3 .. *␤ expecting any of:␤ dotty method or postfix␤»
lizmat r: say v7.1.2 ~~ v.6.2.3 .. * 11:58
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Confused␤at /tmp/NDCaSxb4HE:1␤------> say v7.1.2 ~~ v.⏏6.2.3 .. *␤ expecting any of:␤ dotty method or postfix␤»
lizmat r: say v7.1.2 ~~ v6.2.3 .. *
camelia rakudo 25a0b3: OUTPUT«True␤»
lizmat r: say v7.1.2 ~~ v6.2.3+
camelia rakudo 25a0b3: OUTPUT«True␤»
lizmat it *is* when you enter version numbers right, sorry for the noise 11:58
r: say v6.1.2 ~~ v6.2.3+
camelia rakudo 25a0b3: OUTPUT«False␤»
lizmat r: say v6.2.3+ ~~ v7.1.2 11:59
camelia rakudo 25a0b3: OUTPUT«False␤»
lizmat hmmmm... 12:04
Juerd Postfix +? 12:05
masak Juerd: no, just part of the v term.
yoleaux 01:26Z <colomon> masak: github.com/masak/html-template/pull/6 .....
colomon \o
masak hai, #perl6
masak looks at le pull request
moritz oh hai masak 12:05
masak colomon++: merged. 12:06
brrt i just read the most daring idea ever 12:13
well, not the most daring idea
but in the context it is pretty daring 12:14
the idea was: python won't implement TCO; maybe we can get away from the 1000-stack-frames-limit if we compress each stack frame
masak I see langs/implementations aiming for TCO all the time, and repeatedly failing. 12:17
seems to me it's highly nontrivial, and affects lots of other parts.
moritz TCO? 12:21
total cost of ownership?
brrt tail call optimisation 12:22
tail call elimination
timotimo thanks, captain obvious 12:23
brrt you're welcome :-) 12:24
timotimo :P
brrt TCO is not that hard if you're scheme 12:29
timotimo in what areas is perl6 unicode handling as implemented by rakudo noticably behind perl5 or other languages? 12:37
moritz for example you can't get the Unicode name and properties for a character in Rakudo 12:38
there's no support for normalizaiton
timotimo ah, i wondered about the first one already; is the way to do that already specced? 12:40
there's also matching on unicode properties for regexes; is that in rakudo already? 12:41
moritz yes 12:42
timotimo neat. 12:43
this is another point where i catch myself thinking "how hard could it be?" :D 12:44
pmurias masak: one big reason why Guido refuses to add TCO to python is that is screws up stack traces 12:46
[Coke] just realized he "knows" lizmat from a mailing list somewhere. huh. 12:52
nwc10 and releated? blog.headius.com/2013/05/on-languag...d-way.html -- Python has similar structural complexities to Ruby and adds in the additional complexity of an introspectable call stack. 13:01
masak I once read an interesting presentation (unfortunately I don't recall the URL) which basically said "tail-call elimination is essential for general object orientation". 13:11
don't remember how solid the argument was for that, either.
dalek p: 2a557de | (Pawel Murias)++ | docs/serialization_format.markdown:
Document invocation spec in the serialization format description.
 13:12
masak I did find this when I searched, though: www.eighty-twenty.org/index.cgi/tec...11001.html 13:13
mathw masak: I'd say tail-call elimination is essential for functional programming, not so sure about imperative OOP. 13:16
masak moritz: yeah, well, that was the thing. it argued for that.
I do wish I found that presentation. 13:17
heh, the above article refers to www.cs.utexas.edu/~wcook/Drafts/2009/essay.pdf which has been up for discussion here before. :)
hah! found it! www.ccs.neu.edu/home/matthias/Prese...op2004.pdf 13:21
I don't have time to read it properly now; teaching.
dalek kudo-js: c8b14e1 | (Pawel Murias)++ | / (2 files):
Begin implementing ops required for multimethods.
 13:28
kudo-js: e3041bd | (Pawel Murias)++ | / (2 files):
Update to the new serialization format. Disable 2 failing tests.
timotimo **@foo should fail early, too, right? 13:32
grondilu rn: gist.github.com/grondilu/5662919 13:56
camelia niecza v24-51-g009f999: OUTPUT«Unhandled exception: Excess arguments to sears-haack, used 2 of 3 positionals␤ at /tmp/II6EUbXOuH line 0 (sears-haack @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1132 (ANON @ 6) ␤ at /tmp/II6EUbXOuH line 4 (simpsons @ 6) ␤ at /tmp/II6EUbXOuH li…
..rakudo 25a0b3: OUTPUT«Too many positional parameters passed; got 3 but expected 2␤ in sub sears-haack at /tmp/K1V9eOtgJ7:17␤ in sub CURRIED at src/gen/CORE.setting:2069␤ in sub simpsons at /tmp/K1V9eOtgJ7:4␤ in block at /tmp/K1V9eOtgJ7:19␤␤»
masak it's in slide 59, stated very clearly after a motivating example: "Object-Oriented Programming in languages that don't require tail-call optimizations makes no sense." 13:58
I still don't know if the example convinces me of this; will have to think about it a bit while I walk to the train.
grondilu rn: sub sh($x, $r) { $r*(4*$x*(1-$x))**(3/4) }; say &sh.assuming(*, 1)(.5) 13:59
camelia niecza v24-51-g009f999: OUTPUT«Unhandled exception: Excess arguments to sh, used 2 of 3 positionals␤ at /tmp/daC8yIBSFe line 0 (sh @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1132 (ANON @ 6) ␤ at /tmp/daC8yIBSFe line 1 (mainline @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting…
..rakudo 25a0b3: OUTPUT«Too many positional parameters passed; got 3 but expected 2␤ in sub sh at /tmp/S6KRP977wT:1␤ in sub CURRIED at src/gen/CORE.setting:2069␤ in block at /tmp/S6KRP977wT:1␤␤»
grondilu rn: sub plus($a, $b) { $a + $b }; say &plus.assuming(*, 1)(.5) 14:00
camelia niecza v24-51-g009f999: OUTPUT«Unhandled exception: Excess arguments to plus, used 2 of 3 positionals␤ at /tmp/vx1HqoFTrh line 0 (plus @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1132 (ANON @ 6) ␤ at /tmp/vx1HqoFTrh line 1 (mainline @ 5) ␤ at /home/p6eval/niecza/lib/CORE.set…
..rakudo 25a0b3: OUTPUT«Too many positional parameters passed; got 3 but expected 2␤ in sub plus at /tmp/uRoZ7k2AIw:1␤ in sub CURRIED at src/gen/CORE.setting:2069␤ in block at /tmp/uRoZ7k2AIw:1␤␤»
grondilu what I am missing here?
pugs: sub plus($a, $b) { $a + $b }; say &plus.assuming(*, 1)(.5) 14:01
camelia pugs: OUTPUT«Inf␤»
moritz grondilu: seems like none of them treat * specially in arguments to assuming
grondilu didn't this use to work? 14:02
moritz ie they pass (*, 1, 5) to &plus, and complain that it's one argument too many
I'm pretty sure rakudo never did that
masak on the surface, the argument reads to me like "Java won't let me define cons lists with tail recursion elimination, boo-hoo!" I fail to see the link to "OOP [without TCO] makes no sense".
moritz is it even specced?
[Coke] argh. I have an installed perl6. if I build a different perl6 using the nqp from that installed perl6, it seems to screw up my precompilations. :| 14:03
moritz masak: I find the whole argument "$thing doesn't make sense because it misses an optimization" not very convincing at all
[Coke] will have to track that down to make sure that's what is really happening.
moritz masak: it only starts to be convencing when $thing is generally unusably slow without that optimization, and I'd argue that OO languages don't generally fall into that trap 14:04
masak I really really like the quote on the last slide, however.
grondilu moritz: there is an example in rosettacode.org/wiki/Currying#Perl_6
rn: sub f($a, $b) { $a - $b } 14:05
say .(1) for &f.assuming(0), &f.assuming(*, 0)
camelia rakudo 25a0b3: ( no output )
..niecza v24-51-g009f999: OUTPUT«Potential difficulties:␤ &f is declared but not used at /tmp/IeDtQKP0Ui line 1:␤------> sub f⏏($a, $b) { $a - $b }␤␤»
grondilu rn: rn: sub f($a, $b) { $a - $b } ; say .(1) for &f.assuming(0), &f.assuming(*, 0)
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/6aGi2v1Ij2:1␤------> rn⏏: sub f($a, $b) { $a - $b } ; say .(1) ␤ expecting any of:␤ argument list␤ postfix␤ infix stopper␤ infix or meta-infix␤ state…
..niecza v24-51-g009f999: OUTPUT«-1␤Unhandled exception: Excess arguments to f, used 2 of 3 positionals␤ at /tmp/m4XRCJofPt line 0 (f @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1132 (ANON @ 6) ␤ at /tmp/m4XRCJofPt line 1 (mainline @ 10) ␤ at /home/p6eval/niecza/lib/CORE.setti…
colomon rn: rn: sub f($a, $b) { $a - $b } ; say .(1) for &f.assuming(0)
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/o1EHAfJYwp:1␤------> rn⏏: sub f($a, $b) { $a - $b } ; say .(1) ␤ expecting any of:␤ argument list␤ postfix␤ infix stopper␤ infix or meta-infix␤ state…
..niecza v24-51-g009f999: OUTPUT«-1␤»
grondilu rn: sub f($a, $b) { $a - $b } ; say .(1) for &f.assuming(0), &f.assuming(*, 0)
camelia niecza v24-51-g009f999: OUTPUT«-1␤Unhandled exception: Excess arguments to f, used 2 of 3 positionals␤ at /tmp/EOOPTW0wpc line 0 (f @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1132 (ANON @ 6) ␤ at /tmp/EOOPTW0wpc line 1 (mainline @ 8) ␤ at /home/p6eval/niecza/lib/CORE.settin… 14:06
..rakudo 25a0b3: OUTPUT«-1␤Too many positional parameters passed; got 3 but expected 2␤ in sub f at /tmp/wXOQno1l6M:1␤ in sub CURRIED at src/gen/CORE.setting:2069␤ in block at /tmp/wXOQno1l6M:1␤␤»
moritz grondilu: that is suspicious. It starts with "all Callable", but in Rakudo method assuming is in Routine
method assuming($r: *@curried_pos, *%curried_named) {
return sub CURRIED (*@pos, *%named) {
$r(|@curried_pos, |@pos, |%curried_named, |%named)
}
}
that's rakudo's implementation
no special-casing of * whatsoever 14:07
colomon method assuming(|cap) { sub (|new) { self.(|cap, |new) } }
(that's niecza's)
masak moritz: why does the source code name it CURRIED ?
that just looks like an unused name to me.
grondilu I did write this example in RC. I thought I had discussed it here before. And I also tried it (I guess) 14:08
moritz masak: so that introspecting the return value gives an idea what's going on
masak oh!
fairy nuff
if there were another way to annotate a sub, I'd almost want to use that way instead. 14:09
moritz grondilu: irclog.perlgeek.de/search.pl?channe...q=assuming looks like you tried in April, and it didn't work back then either
masak so that in a stack trace, it said "a sub that was primed on line X, file Y"
timotimo hm, would pandabrew re-install all installed modules when i build a new rakudo? 14:11
dalek kudo-js: 5755640 | (Pawel Murias)++ | run (2 files):
Implement nqp::flip, pass test 59 again.
kudo-js: fa579be | (Pawel Murias)++ | run (2 files):
Pass test 44 again.
tadzik timotimo: it should, yes 14:12
timotimo: (I don't know if that works, though :D)
if not, then I'll fix that by YAPC::NA
but first I need to fix panda itself
starting with daxim's pet bug 14:13
[Coke] my $r; say $r{3}:exists;
r: my $r; say $r{3}:exists;
camelia rakudo 25a0b3: OUTPUT«No such method 'exists' for invocant of type 'Any'␤ in method postcircumfix:<{ }> at src/gen/CORE.setting:1663␤ in method postcircumfix:<{ }> at src/gen/CORE.setting:1642␤ in block at /tmp/Mon6Dc82_4:1␤␤»
daxim :-[∿]
[Coke] r: my $r = Hash.new(); say $r{3}:exists;
camelia rakudo 25a0b3: OUTPUT«False␤» 14:14
[Coke] r: my $r = Hash.new(); my $n = 3; say $r{$n}:exists;
camelia rakudo 25a0b3: OUTPUT«False␤»
[Coke] r: my $r = Hash.new(); my $n = 3; say "eek" if !$r{$n}:exists; 14:15
camelia rakudo 25a0b3: OUTPUT«Unexpected named parameter 'exists' passed␤ in sub prefix:<!> at src/gen/CORE.setting:2734␤ in block at /tmp/5ZWRDN1yxD:1␤␤»
[Coke] r: my $r = Hash.new(); my $n = 3; say "eek" if ! $r{$n}:exists;
camelia rakudo 25a0b3: OUTPUT«Unexpected named parameter 'exists' passed␤ in sub prefix:<!> at src/gen/CORE.setting:2734␤ in block at /tmp/t9r2sbslv5:1␤␤» 14:15
[Coke] rakudobug?
[Coke] r: my $r = Hash.new(); say "eek" if ! $r{3}:exists; 14:15
camelia rakudo 25a0b3: OUTPUT«Unexpected named parameter 'exists' passed␤ in sub prefix:<!> at src/gen/CORE.setting:2734␤ in block at /tmp/xJy6R7yxgO:1␤␤»
timotimo ! is too tight? use not instead i guess 14:16
[Coke] r: my $r = Hash.new(); say "eek" if ! ($r{3}:exists);
camelia rakudo 25a0b3: OUTPUT«eek␤»
timotimo std: my $r = Hash.new(); say "eek" if ! $r{3}:exists;
camelia std 0336087: OUTPUT«ok 00:00 45m␤»
timotimo yeah, that's not helpful :( 14:17
moritz [Coke]: IMHO yes. If Any reponds to {}, it should also respond to {}:exists 14:19
masak decommutes
timotimo tadzik: would you accept a patch that makes panda not explode when a module has no lib/ folder? 14:51
tadzik timotimo: oh yes 15:11
does it explode?
timotimo yeah, there's a file::find that will complain, because lib doesn't exist 15:14
i just added an if 'lib'.IO.d before it
tadzik oh hm 15:15
so _that's_ what the check was for, at the beginning of build() :D
we need a test for that
[Coke] I wonder if that iospec branch merge is relying on destructors to clean up resources. 15:55
moritz fwiw the bool.t failure has been there on and off for years 16:22
sometimes it disappears, sometimes it reappears 16:23
and usually the things that trigger it have nothing to do with the underlying problem
moritz it's probably some weird memory corruption, and then adding a built-in changes the memory layout so the the bug is triggered, or something 16:23
Util r: class AA { multi method new { AA.bless: * }}; class BB is AA {}; say BB.new.WHAT; 16:25
camelia rakudo 25a0b3: OUTPUT«(AA)␤»
Util r: class AA { multi method new { ::?CLASS.bless: * }}; class BB is AA {}; say BB.new.WHAT;
camelia rakudo 25a0b3: OUTPUT«(AA)␤»
Util r: class AA { multi method new { self.bless: * }}; class BB is AA {}; say BB.new.WHAT;
camelia rakudo 25a0b3: OUTPUT«(BB)␤»
Util r: class AA { multi method new { $?CLASS.bless: * }}; class BB is AA {}; say BB.new.WHAT;
camelia rakudo 25a0b3: OUTPUT«(AA)␤»
Util Shouldn't the 4th version say BB like the 3rd?
jnthn No, it's not virtual. 16:26
yoleaux 11:35Z <lizmat> jnthn: that there is already a fix for .exists on typed hashes: it's called github.com/rakudo/rakudo/pull/147
jnthn Use self for that.
Util jnthn: Thanks!
dalek rl6-bench: 21eecd0 | (Geoffrey Broadwell)++ | jqplot/plugins/jqplot.canvasAxisLabelRenderer.min.js:
Add jqPlot plugin needed for rotated axis label rendering
 16:27
jnthn .tell lizmat_ I wonder if the Mu \key shouldn't be TKey \key
yoleaux jnthn: I'll pass your message to lizmat_.
nwc10 it's jnthn! 16:28
jnthn waves hazily 16:29
nwc10 where in the world is jnthn, and is the beer good? 16:30
arnsholt hopes the beer in Austin will be good 16:31
arnsholt s/hopes/hops/ # =) 16:31
jnthn is at home, and somewhat sick
nwc10 oh yay :-(
jnthn Teaching while sick is...fun. 16:32
arnsholt Not too sick for a trans-atlantic flight, I hope?
PerlJam jnthn: drink orange juice, get plenty of rest, take vitamins, take medicine, etc. :)
jnthn Teaching while sick and having somebody give you full marks in the evaluation is a darn miracle.
arnsholt: I highly doubt that is at risk.
I have Thu/Fri with nothing to do but rest and prepare for YAPC::NA.
Tomorrow I got a couple of evening talks to give. 16:33
uh, I didn't write one of them yet. oops.
[Coke] mmhehehe
jnthn And I only slept 3 hours last night. This will be interesting...
arnsholt jnthn: Excellent! I was looking forward to beer with you and masak in Austin =) 16:34
Am looking forward to, even 16:35
masak hello from a train! \o/ 16:39
moritz (wlan on train)++
masak moritz: well, could've been 3G on le mobile phone, too. 16:41
masak rn: my %l = foo => 1, bar => 2; my %r = bar => 3, baz => 4; say %l >>+<< %r 17:01
camelia niecza v24-51-g009f999: OUTPUT«Use of uninitialized value in numeric context␤ at /home/p6eval/niecza/lib/CORE.setting line 1355 (warn @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 294 (Any.Numeric @ 8) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.s…
..rakudo 25a0b3: OUTPUT«use of uninitialized value of type Any in numeric context in block at /tmp/OI858EugnG:1␤␤use of uninitialized value of type Any in numeric context in block at /tmp/OI858EugnG:1␤␤("foo" => 1, "bar" => 5, "baz" => 4).hash␤»
masak I must say I don't need that one often... but when I do, it's *very* nice to have. 17:02
what's with all the warnings, though?
isBEKaml heh, what does >>+<< on hashes mean? do we take in keys too?
masak isBEKaml: do the operation on the values. >>+<< takes the union of the keys; <<+>> takes the intersection. 17:03
isBEKaml: in this case I needed the union.
masak which is why I think the undefinedness warnings are unnecessary. 17:03
I *know* some keys/values are missing. that's why I did >>+<<
masak submits rakudobug 17:04
isBEKaml masak: so, this is one of those cases where both rn got it wrong? (where it's specced ofcourse) 17:05
masak isBEKaml: both rakudo and niecza warn, so that part is wrong.
isBEKaml: niecza warns so much that I can't see if it got the right result. :( 17:06
masak doesn't think warnings should have stack traces
isBEKaml masak: ah, overnieczathingamajigging?
masak fails to understand the question 17:07
please try again :)
isBEKaml masak: so that's what Niecza thinks!
isBEKaml had a funny experience of RT'ing a Mindblown fact only to have Yoda send me a message. (it involves some Star wars characters, though) 17:09
isBEKaml Tweet: twitter.com/MindBlowing/status/339...3208716288 17:11
masak most MindBlowing tweets are in fact mundane, clearly false, or misinformed. :/ 17:19
this one, for example: twitter.com/MindBlowing/status/338...1290575872
reply. "no. also, you fail at genetics. thanks for playing."
isBEKaml masak: I vaguely recall mendel charts. :D 17:21
timotimo "People with the highest IQ stay up late at night because their brains have increased mental stimulation." - wha? 17:28
masak or it could just be a lack of discipline :) 17:32
or susceptibility to flattery :)
or those poor folks who seem to have a 28-hour day cycle.
PerlJam masak, timotimo: clearly the two of you are way too serious today ;)
masak PerlJam: someone is Wrong On The Internet! 17:33
[Coke] masak: Unfortunately, it might be you. 17:35
timotimo .o(mind blown!)
hashtag epic twist
masak [Coke]: hehe. I guess that's a risk I'm willing to take.
timotimo as does everybody on the internet, it seems 17:37
dalek rl6-roast-data: 00ea812 | coke++ | / (4 files):
today (automated commit)
 18:09
[Coke] if niecza ever cleans up, my "how long has it been failing" script will start running faster. :) 18:10
niecza has been dirty for 315 days. pugs has been clean for 11 days. rakudo has been dirty for 24 days.
pmichaud rakudo's dirtiness is still the bool.t thingy? 18:18
colomon [Coke]: believe it or not, we're being paid by the electric company to waste your computing power...
[Coke] pmichaud: yes. 18:19
there's an RT with gory details and backtraces.
pmichaud okay.
[Coke] if someone has the iospec branch someplace, I can see which of those commits triggered it.
pmichaud since I can reproduce it fairly well on feather, I might be able to provide a couple of extra clues 18:20
[Coke] (I know it's not the root cause, but it's triggering something)
dalek osystem: 257e82c | colomon++ | META.list:
Move Test-Junkie over to working fork

Temporarily move Test-Junkie module over to my fork which passes all tests.
[Coke] pmichaud: also, you fixed R*'s daily test the other day. 18:21
it's been clean since then.
pmichaud oh!
so I can do an R* release, perhaps?
that would be really good before next week.
I'm a bit swamped today, but that could be a good $Wed task.
[Coke] github.com/coke/rakudo-star-daily/...odules.log - no fails. 18:22
isBEKaml This is more of a packaging question than anything else - Parrot has not marked 5.2 to be a supported release, should R* be basing itself off a Parrot developmental release? 18:24
Or is that merely a convention on Parrot's part?
[Coke] there is really no difference between those kinds of release.
pmichaud I can certainly bundle 5.3.0 with Star if there's a good reason to do so. 18:25
timotimo pmichaud: if you can get it before saturday, that would be before my hopefully at least minimally great perl6 talk and i could tell people "we just had a new release" rather than "grab the release from 5 months ago with a lot of bugs and outdated specs, or be crazy and spend an hour downloading and compiling three things!" ;)
pmichaud timotimo: if it happens, it'll have to be Wed or Thu
Fri and Sat will be busy for me preparing for YAPC::NA
[Coke] pmichaud: if you bump the req to 5.3.0 it'll run on feather for a few days and you can see if anything errors out. 18:26
timotimo that's fine :)
thanks for doing it
[Coke] ah. so given the timings, I'd say keep whatever is in R*'s repo now.
but I'd bump it soon after for the testing.
pmichaud oh, it's no problem to test against 5.3.0
pmichaud it's not like a PARROT_REVISION bump in NQP or Rakudo 18:26
[Coke] ah. I find the testing horrifically slow, which is why I automated it. :)
pmichaud with star, you just say which version of Parrot you'd like to have bundled. 18:27
masak r: class A {}; my %h{A} = A.new => 2; my %h2{A} = %h >>+>> 40; say %h2 18:27
camelia rakudo 25a0b3: OUTPUT«Nominal type check failed for parameter 'key'; expected A but got Str instead␤ in method STORE_AT_KEY at src/gen/CORE.setting:7031␤ in method STORE at src/gen/CORE.setting:6927␤ in block at /tmp/mYmQfsTMvo:1␤␤»
pmichaud i.e., star doesn't change the minimum required version that NQP wants... star just chooses to supply a version of Parrot in the bundled (which should be at least PARROT_REVISION, but doesn't have to be exactly PARROT_REVISION)
masak how reasonable are my expectations above? I expect that to work. 18:28
isBEKaml pmichaud: I can help building R* this weekend, but no .msiS :-)
masak (i.e. produce '(A.new() => 42).hash' as output)
moritz 18:28
~ .
pmichaud masak: I suspect that hypering doesn't know about typed hash keys 18:29
masak I suspect so too.
pmichaud since it was written before we had them :)
masak but... it's not unreasonable to have that work, is it? :)
pmichaud on its face it doesn't strike me as unreasonable 18:30
masak submits rakudobug
pmichaud the devil may be in the details here, though
I consider it a valid rakudobug for the moment, yes.
[Coke] anyone looking for some low hanging fruit? I can point you at a ticket to write a ticket for. 18:31
to write a *test* for
masak .oO( rakudobug status may be revoked at any future time, with no possibility of appeal ) :P 18:31
isBEKaml Speaking of R*, is there any way I can just download a specific branch? I'm not looking forward to cloning a 200 meg repository. :/ 18:32
moritz isBEKaml: the R* .git is just 14MB
pmichaud isBEKaml: ummmmm.... I have trouble following your question.
the R* repo doesn't actually contain copies of Rakudo, Parrot, or NQP 18:33
isBEKaml moritz: No, I didn't mean that. perl Configure.pl step runs gen-parrot, which actually clones a fresh parrot repo, right?
pmichaud the rakudo/star repo is a toolkit for building R* distributions. It's not the R* distributions themselves.
moritz isBEKaml: no. If you download an R* release, and run the Configure.pl, it uses releases of parrot, nqp and rakudo 18:34
pmichaud --gen-parrot in a star tarball uses the bundled copy of Parrot. No cloning.
isBEKaml pmichaud: I spoke of building an R* package. :-)
pmichaud (you spoke of downloading a branch :) 18:35
masak pmichaud: hm. question is, if I %h1 >>+<< $h2, and the hashes are typed with different constraints, what the result ends up being. same question for <<+>>, really.
(typed hashes and hyperoperators! yay, I found a new corner case!)
pmichaud masak: yes, that's a legitimate question.
isBEKaml pmichaud: so I can just download a branch manually from Github and place it on the source location for R*? That's nice.
pmichaud isBEKaml: when building an R* tarball, it actually downloads a Parrot tarball
it doesn't clone the Parrot repo. 18:36
[Coke] unlike when building rakudo compiler. 18:36
masak pmichaud: I guess the type-theoretical answer is that in one case you'll get the join of the types, and in the other case you'll get the meet of the types :P
pmichaud masak: I'm thinking the practical answer may be DIHWIDT for a while :)
pmichaud (building an R* tarball does clone the module repositories at the moment, but we could undoubtedly change/improve that if we want) 18:37
isBEKaml pmichaud: thanks, if that's the case - it makes me feel better. :-) 18:38
isBEKaml pmichaud: Right, I see that in star's makefile now. I'm alright with cloning individual module repositories since they are all pretty small on their own. 18:40
pmichaud Star wants to be created only from actual releases, so it uses the release tarballs instead of github tags. 18:41
when/if we get to the level that modules have release tarballs, R* will undoubtedly want to use those as well. cloning github repos directly is a bit fragile. 18:42
(in terms of version maintenance and stability)
isBEKaml pmichaud: that's quite some distance away since it requires that we have a cpan like storage facility. 18:43
timotimo could the setup be changed to clone only with --shallow when installing an R* release?
moritz timotimo: there's no cloning when installing an R* release.
timotimo: it bundles everything
fwiw I have a few modules that I'd release, if we had the tools to make it not too painful 18:44
(Math::RungeKutta and JSON for example)
timotimo ah, of course
JSON isn't in the R*?
moritz it is 18:45
timotimo oh, you mean tarball release, duh
pmichaud isBEKaml: it was less of a distance when github offered file downloads. Yes, it's a bit farther away. OTOH, there are groups that are working on being able to upload Perl 6 modules into CPAN 18:48
so, we already have a cpan-like storage facility, it's called "CPAN" :)
isBEKaml pmichaud: :-)
PerlJam pmichaud: who is working on being about to upload Perl 6 modules to CPAN? 18:49
s/about/able/
nwc10 www.cpan.org/ -- "
CPAN
Comprehensive Perl Archive Network"
naughty copy-paste 18:50
pmichaud PerlJam: I'm not sure I'm at liberty to disclose that atm.
I don't know how public the effort is.
masak so, I've been playing around with a Perl 6 solution to xkcd.com/287/ tonight.
here's my code. it works. gist.github.com/masak/5664729
nwc10 not sure if there's much Perl 4 on CPAN. 18:51
PerlJam nwc10: or Perls 3, 2, and 1
masak I think I can speed it up somewhat by pruning branches. trying that now.
PerlJam nwc10: though keep in mind that CPAN didn't exist until Perl 5
masak hah! down from 183 seconds to 19 seconds! pruning win! 18:53
masak I'd be very happy to receive comments and suggestions for improvements on the program. 18:55
pmichaud var() seems keywordish to me.
as in: there might be a name conflict or confusion somewhere. 18:56
masak ...with what?
pmichaud I don't have a better alternative.
I think I initially worried about a conflict with "val".
masak down to 10.5 seconds. (fixed a logical bug.)
pmichaud or one of those other recent spec changes that I haven't caught up with yet.
masak I like "var()". as long as I'm not overwriting anything from the setting, I'm OK with it. 18:57
pmichaud yeah, there's a val() function I was thining of. 18:58
*thinking
masak I wrote this program backwards. started by writing out lines 82..89, and then asking myself "what do I need for that to work?" 18:59
pmichaud masak: yeah, that's what I thought when I first read it.
"masak just wanted the bottom part to work"
masak :) 19:00
PerlJam masak: I have no comments on the code other than I like how readable it is.
pmichaud I wonder if it becomes more interesting if you can do %V<mixed fruit> instead of var("mixed fruit") somehow.
then it looks more like a variable. :)
masak (except that that's not how you index a single hash value?)
pmichaud (yeah, except that :) 19:01
or even V.new('mixed fruit')
masak maybe I can get rid of the Variable class altogether... trying. 19:02
TimToady for generalized constraint programming it probably wants a declarator of some sort
yoleaux 26 May 2013 06:53Z <[Coke]> TimToady: if we should switch the names of FIRST and START, maybe.
pmichaud yeah, maybe CV (for "constrained variable") could be the class name or something.
anyway, all I have are cosmetic suggestions at the moment. 19:03
I think it's cool you were able to do this.
TimToady I think FIRST is fine, but maybe we could rename START to ONCE 19:04
timotimo masak: i'd like to suggest sym instead of var if var turns out not to like
pmichaud sym might be ickier -- puns with category:sym<foo> perhaps, and the <sym> regex rule. 19:05
timotimo alternatively, there could be a prefix operator
masak shorter version: gist.github.com/masak/5664729 19:06
[Coke] r: my @a = gather for 1..100 { say $_; take $_}; 19:07
camelia rakudo 25a0b3: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤81␤82␤83􏿽xE2
[Coke] I assume I'm missing some reason why that isn't lazy?
timotimo r: sub prefix:<The> ($name) is tighter(&infix:<+>) { "'$name'" }; say (The 'mixed fruit' + The 'french fries'); 19:08
pmichaud array assignment is always eager.
camelia rakudo 25a0b3: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏'mixed fruit'' (indicated by ⏏)␤ in method Numeric at src/gen/CORE.setting:10063␤ in sub infix:<+> at src/gen/CORE.setting:2983␤ in sub infix:<+> at src/gen/CORE.setting:298…
TimToady r: my @a = gather for 1..100 { say take $_};
camelia rakudo 25a0b3: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤81␤82␤83􏿽xE2
[Coke] pmichaud:danke.
timotimo r: sub prefix:<The> ($name) is tighter(&infix:<~>) { "'$name'" }; say (The 'mixed fruit' ~ The 'french fries');
TimToady r: my @a := gather for 1..100 { say take $_};
camelia rakudo 25a0b3: OUTPUT«'mixed fruit''french fries'␤»
rakudo 25a0b3: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤81␤82␤83􏿽xE2
TimToady apparently := is also being eager 19:09
pmichaud also, I think that moritz made 'for' eager.
TimToady boo hiss
pmichaud r: my @a := gather (1..100).map( { say take $_ } );
camelia rakudo 25a0b3: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤81␤82␤83􏿽xE2
pmichaud r: my @a := gather (1..100).map( { say take $_ } ); 1;
camelia rakudo 25a0b3: ( no output )
pmichaud also, displaying @a makes it eager.
flussence rn: [+] $_ X- .pop given 'math`'.ords # irclog.perlgeek.de/perl6/2013-05-26#i_7117448 19:10
TimToady r: my @a := gather for 1..100 { say take $_}; 1
camelia rakudo 25a0b3, niecza v24-51-g009f999: ( no output )
rakudo 25a0b3: ( no output )
TimToady whew! :)
[Coke] moritz had changed for, then changed it back because pretty much report from the wild was "eew, this is horrible"
masak down to 5 seconds.
flussence rn: say [+] $_ X- .pop given 'math`'.ords
camelia rakudo 25a0b3, niecza v24-51-g009f999: OUTPUT«42␤»
pmichaud okay, I didn't see the changeback.
moritz++
masak and lost one of the solutions, too :(
flussence (I can make it shorter, but that line is cuter)
[Coke] pmichaud: ... I don't know which way it is now. ;)
pmichaud there's still some nastiness involving sink context and array variables, I suspect 19:11
masak ah, logical bug again. :)
pmichaud so having "my @a :=" as the last line in the mainline may be giving it some forced eagerness.
[Coke] r: my $a = gather for 1..100 { say $_; take $_};
camelia rakudo 25a0b3: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤81␤82␤83􏿽xE2
[Coke] r: my $a := gather for 1..100 { say $_; take $_};
camelia rakudo 25a0b3: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤81␤82␤83􏿽xE2
[Coke] so, it's not just array assignment. 19:12
pmichaud add a "1;" there.
[Coke] r: my $a := gather for 1..100 { say $_; take $_}; 1;
camelia rakudo 25a0b3: ( no output )
TimToady or read the backlog
[Coke] EW!
so, for is the 'ewwwwww' way, now.
pmichaud it's because the result of the statement list is $a, which then gets .sink called on it.
[Coke] ew.
pmichaud (which I still disagree with, but I haven't had tuits to be able to review the sink changes and fix them)
TimToady note that gather technically puts its for into sink context as well, but the take should suspend that 19:13
pmichaud right, the .sink in this case is being called on the List containing the GatherIter
timotimo i'm kind of sad that ::CALLER:<$?LINE> doesn't seem to work :( 19:19
moritz timotimo: callframe().line should
timotimo thanks :) 19:20
it even gives the file, that's great :)
masak current solution down to 7 seconds.
masak suggestion: rename var() to unknown() 19:21
timotimo masak: your script doesn't run on niecza, because n doesn't seem to like adding multis to +, * and == 19:22
flussence r: say 'a' ~~ /(.)/ # just doing this to see if the regex quote brackets work on my screen...
camelia rakudo 25a0b3: OUTPUT«｢a｣␤ 0 => ｢a｣␤␤»
masak timotimo: that's niecza's problem. :) 19:23
timotimo inded 19:24
pmichaud or unk() if you want shorter. 19:27
which is surprisingly close to "thunk" :-)
afk, kid pickups
[Coke] I have a naive solution to the masak challenge that pretty much times out due to using the combine() from RC on a list of 31 taken 4 at a time. 19:28
lizmat
.oO(second time someone mentions unk today)
 19:29
masak here's a corresponding MiniZink model: www.hakank.org/minizinc/xkcd.mzn
[Coke] lizmat: you were talking about how you see Any and Mu earlier. I just see them as points in the type hierarchy to control junctions. "use this one to autothread junctions" "use this one to treat junctions as self contained things with no magic powers" 19:31
lizmat yes, I was taught this by jnthn at PLPW
I didn't know about that property of Mu when I was thinking about using Mu 19:32
[Coke] ah. alrightee then.
masak yes, I'm happy with my solution now.
I didn't really expect to arrive at one that runs in < 10 seconds on Rakudo.
even though the searchspace is not enormous.
Util #ps time 19:34
moritz doesn't have anything to conribute to #ps these days 19:36
lizmat is just lurking
[Coke] adding the gather/take to rosettacode.org/wiki/Combinations#Perl_6 seems to have slowed it down a bit. 19:40
er, a LOT. 19:42
[Coke] huh. or something on feather was killing the CPU there. 19:47
lizmat .tell jnthn I did the suggested change and it spectested ok. I also removed the ::?CLASS:D: as the other methods taking a key didn't have that either 19:58
yoleaux lizmat: I'll pass your message to jnthn.
dalek kudo-star-daily: ef7b1f1 | coke++ | log/ (5 files):
today (automated commit)
 20:00
[Coke] no change on R*
lizmat has changed github.com/rakudo/rakudo/pull/156 to only change the API of Hash.new and not touch the API of hash() 20:13
lizmat has removed some Mu's from github.com/rakudo/rakudo/pull/156 , because she now realizes that was not really a smart thing to do (even though it spectested ok) 20:14
lizmat *158 20:14
FROGGS o/ 20:15
lizmat FROGGS!
FROGGS :o)
lizmat also killed a Mu in github.com/rakudo/rakudo/pull/147 20:16
FROGGS pmichaud / [Coke]: my home-directory on feather has a checked-out iobranch merge, and some local changes... 20:21
at this state, the bug hides if you comment out a single line of the test file
the commented-out lines of IO/Spec/Unix.pm will trigger the bug too if commented in (with an unchanged test file) 20:22
lizmat started github.com/perl6/specs/issues/45 (use Foo:version<...>, how to deal with edge cases?), hoping for a resolution in Austin, TX 20:25
lizmat seems we have some bad net weather? 20:34
FROGGS might be just me on a train 20:38
dalek ecs: ff6e953 | (Elizabeth Mattijsen)++ | S11-modules.pod:
Revert "Revert "Remove line the was left behind from a previous edit""

This reverts commit 546a9c4322e537f98a71d99f0dc473532b416e10.		 20:56
jnthn ... 20:57
yoleaux 19:58Z <lizmat> jnthn: I did the suggested change and it spectested ok. I also removed the ::?CLASS:D: as the other methods taking a key didn't have that either
jnthn Somebody in one of my Git classes once asked, "can you revert a revert". Now I've seen it happen in the ild...
ajr_ I assume dalek is a bot of some kind, perhaps tracking updates?
jnthn *wild
lizmat jnthn: I found that I actually had reverted a change that shouldn't have been (afaik)
tadzik good evening #perl6 20:58
lizmat tadzik!
jnthn ajr_: Tracks commits, yet. And with a link so you can code review what was commat
o/ tadzik
*ye
*yes
lizmat how did it go today / yesterday?
masak jnthn: yes, but a revert of a revert is just a cherry-pick of the original commit. that's why cherry-picks are covariant and reverts are contravariant ;)
jnthn argh, can't type
masak tadzik! \o/
jnthn masak: Huh? Does it actually do it that way, or the effect is just the same?
ajr_ Np. I got the gist. :-)*
masak tadzik: the two Ciechans are now safely in my cupboard, to be consumed on some hot summer's day. :) 20:59
jnthn: the effect is just the same.
jnthn Oh, phew :) 20:59
masak jnthn: I'm pretty sure enough information is lost that it doesn't actually go back to the original commit ;)
jnthn Right, I was pretty sure it didn't actually store enough to implement it that way.
masak jnthn: just pointing out that you could've just the re-cherry-picked the original commit.
tadzik masak: do give me a heads-up when you consume them, I'm curious about their contents
masak .oO( I have a feeling it's beer... ) :P 21:00
tadzik: will do.
tadzik good, good. Let the beer flow through you.
masak I'll make sure to invite jnthn, as per agreement. 21:01
timotimo tadzik: i've just made a pullrequest for the test for missing lib/ and the fix i made 21:16
tadzik ossum
timotimo even more slides procrastinated away! ;)
tadzik before I merge pull requests, I like to fetch them and test locally 21:18
github could use a textbox with "git clone this.and.that to test it locally"
so I don't have to click on 'timo', look through your repos, find panda, find branches...
dalek nda: 99ecd04 | (Timo Paulssen)++ | / (2 files):
don't explode when no lib folder exists +test
 21:20
nda: 1b27a1d | tadzik++ | / (2 files):
Merge branch 'accept_missing_lib'
tadzik timotimo++ # thanks! 21:21
having contributors is awesome :)
masak it is. contributors++
timotimo if you write it like that, masak, it seems like a volunteerment ;) 21:24
masak hides behind a git branch
timotimo so, i was hoping to use callframe() to get the filename and line of where i was called from 21:26
turns out it goes through the core setting for quite a few levels m)
timotimo r: say callframe(5).file; say callframe(5).file; 21:32
camelia rakudo 25a0b3: OUTPUT«src/stage2/gen/NQPHLL.nqp␤src/stage2/gen/NQPHLL.nqp␤»
timotimo oh no! 21:33
i'm introducing new callframes by using "first" to find the first callframe outside of my lib >_<
do those frames get invalidated when one gets popped off, but not down to the one i'm holding a reference to? 21:34
jnthn Slides from my talk at Polish Perl Workshop: jnthn.net/papers/2013-plpw-rakudo-jvm.pdf 21:37
lizmat jnthn++ 21:40
good to see it again, it was a lot to take in :-)
masak oh right, slides.
I'll try to get mine published tomorrow.
lizmat given a file A.pm with: module A:version<v1> { sub a is export { say "a" } }; module B:version<v1> { sub b is export { say "b" } } 21:47
lizmat use A; a; b; import B # currently fails because it tries to import "b", whereas the "use A" already did that also 21:48
even though they are the same sub 21:49
lizmat is the fact that "use A" exports the subs from module B a rakudobug, or is that a case of "don't do that" ? 21:50
jnthn Well, you're importing the stuff from EXPORT::DEFAULT
And "is export" installs the things in there
lizmat Feels to me that we should maybe disallow more than one module / class etc. at the top level of a UNIT to prevent these type of situations? 21:51
jnthn Nah.
It's common to do that. Especially if one of them is a lexical class that you're using as an implementation detail
lizmat I'm not against that
that's lexical inside, and hidden from the outside
it's just that there is a difference between "use A" and "import A" in the case of 2 top level modules with exportable subs 21:52
jnthn I thought we didn't complain about importing the same things twice, though...
r: use Test; use Test;
camelia rakudo 25a0b3: OUTPUT«===SORRY!===␤Cannot import the following symbols from Test, because they already exist in this lexical scope: &skip_rest, &diag, &done_testing, &done␤at /tmp/zuyxQPOgEB:1␤------> use Test; use Test⏏;␤»
jnthn Ah, hm.
lizmat so maybe that's the rakudobug ? 21:53
jnthn I'm a bit surprised it complains there.
Especially as I thought I'd fixed that in the past...
lizmat ok, then I'll submit a rakudobug for this and we can look at this later
is that a plan?
jnthn Sure 21:54
lizmat submits rakudobug 21:55
goodnight #perl6! and sleep&
timotimo jnthn: how much work do you figure will it take for Rakudo::Debugger to run on the jvm? from my point of view it should range from "works immediately" to "an evening's worth of work" (of course assuming a sufficiently matured rakudo-jvm) 21:58
jnthn timotimo: Probably O(day), possible O(evening), but yeah, needs a sufficiently matured rakudo-jvm :) 21:59
masak .oO( Society Against The Abuse Of Ordo Notation ) 22:06
dalek kudo/nom: be60ebe | dagurval++ | src/core/IO.pm:
fix order of directory/volume used in dir()
kudo/nom: 2f8ebe1 | (Brent Laabs)++ | t/spectest.data:
add S02-names/bare-sigil.t to spectest
rakudo/nom: a418e53 | dagurval++ | src/core/Temporal.pm:
labster that was supposed to be: Add IO::Handle meth: gist, perl, path (now returns IO::Path), IO (a no-op) 22:09
Juerd What is the user p6eval on feather1? Evalbots must 15399 p6eval 20 0 304m 274m 1928 R 0,4 23,3 1:37.04 perl6 22:32
12384 sergot 20 0 499m 182m 684 D 0,0 15,5 4:32.28 perl6
Arghhhh
Lag + paste fail
Evalbots must be on feather3, where they can cause little harm
So I'd like to learn what p6eval is on feather1; It's causing problems now.
sorear Juerd: iirc feather3 doesn't have enough RAM to build rakudo, so we have a cronjob on feather1 which builds rakudo and scps it to feather3 22:36
all user evaluations are done on feather3
labster *sigh*, now people are trying to pick fights on the YAPC email list. 22:38
sorear labster: is that in the movie night thread I just deleted without reading? 22:39
labster one of them. title is "dha is the catalog!" 22:40
jnthn I've yet to see a mail on that list that is actually worthwhile. 22:42
labster Me neither. 22:44
jnthn wonders how you un-sub... 22:48
ah, found it
(just follow listinfo link at bottom of posts) 22:49
masak oh, a good idea. 22:54
masak follows suit
here's to staunchly refusing to let the YAPC::NA mailing list ruin the YAPC::NA experience! :D 22:55
'night, #perl6 22:56
sorear night masak 23:00
jnthn 'night from me too 23:13

Powered by Cro and the Raku® Programming Language.

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

Thank you!