»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:12 llfourn joined 00:16 llfourn left 00:18 daxim joined 00:21 laouji joined 00:27 perturbation left 00:31 raiph left 00:39 laouji left, laouji joined 00:44 laouji left 00:46 laouji joined
Sgeo As someone who has limited experience with Perl but much more experience with Python and Haskell, what's interesting about Perl6? 00:47
augment: benign parasitic. supersede: deadly parasitic 00:50
ShimmerFairy Not that I can speak much to comparing P6 to P5 (since I only know 6), but the design of Perl 6 is a bit more "modern" than P5; for example, when I initially tried using P5, I got tripped up on an introductory Qt tutorial because of -> for methods, among other things.
ugexe you can sum that up as MONKEY-TYPING in general, and thats hardly new to perl6 00:51
Sgeo I have no idea what parasitic means nor why PErl6 has something called deadly parasitic 00:52
ugexe i thought those were your words 00:53
ShimmerFairy Sgeo: "augment" means "let me give this class some new features", "supersede" means "I can do a better job than you, move over!" :)
Sgeo ugexe, raw.githubusercontent.com/perl6/mu...tsheet.txt 00:54
timotimo yo Sgeo 00:55
Sgeo hi
timotimo for a moment i was surprised you didn't know anything about perl6, since i've seen you in here for so long ... then i noticed i know you from a differen tchannel %)
ugexe yea i dont know what those mean either Sgeo 00:59
Sgeo timotimo, heh. I don't really talk much on that channel though
ShimmerFairy "parasitic" I assume refers to the fact that, like any good parasite, "augment" and "supersede" attach to and affect classes that didn't really ask for it :) 01:02
ugexe sounds like a role 01:03
ShimmerFairy It's not like a role or mixin though, because it affects the class in question for the entire scope (and maaaaybe outer scopes too, but I can't recall) 01:04
ugexe the benign/deadly parasitic are referring to scopes though. so i guess maybe its scope is dependent on the parasites host, but doesnt do much to explain benign/deadly 01:07
ShimmerFairy No, benign/deadly is surely a description of what the keywords do ("augment" changes, "supersede" replaces) 01:08
ugexe its grouped directly under my, our, has, anon, and state under a header titled scope. if thats what it means then its organization is confusing to me 01:09
ShimmerFairy It's because 'augment' and 'supersede' occupy the same position as those other keywords in code, though admittedly they're the odder ones in the collection :) 01:11
01:15 BenGoldberg joined
ShimmerFairy To be fair, 'state' doesn't have much to do with scoping either (it's more about having the variable persist in its current scope) 01:16
Sgeo m: my $chapter = 3; undefine $chapter; say defined $a; 01:17
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/vzsvMQQ_g7␤Variable '$a' is not declared␤at /tmp/vzsvMQQ_g7:1␤------> 3ter = 3; undefine $chapter; say defined 7⏏5$a;␤»
Sgeo tablets.perl6.org/tablet-3-variables.html
:/
Starting to think that this tablet is garbage
timotimo the tablets aren't the most up to date resource, sadly 01:18
ShimmerFairy That should probably say 'chapter' instead of 'a' 01:19
timotimo you may want doc.perl6.org instead
and i believe the difference between augment and supercede is that augment will just augment a class whereas supercede will supercede it
ShimmerFairy And yes, the tablets are a pretty old resource, I wouldn't recommend them really. 01:20
Sgeo ty
Maybe they shouldn't be mentioned, or noted as an old resource, on perl6.org
Oh there is a disclaimer but it applies to all resources on the page 01:21
perl6.org/documentation/
ugexe we were discussing the descriptions "benign parasitic" and "deadly parasitic" as related to what augment and supercede do 01:22
timotimo oh 01:23
ShimmerFairy which, as I understand it, is in relation to what augment and supersede do, specifically :)
timotimo well, in "benign parasitic", the host survives, in "deadly parasitic", it doesn't
dalek kudo-star-daily: be23065 | coke++ | log/ (8 files):
today (automated commit)
 01:24
Sgeo What does ~~ :e mean, what kind of thing is :e, a pair? How does smartmatch work against a pair?
ugexe existence adverb on a file 01:25
Sgeo How does smartmatch work on that?
timotimo :e is short for e => True, smart matchign against that calls the method .e and checks if the result is True 01:26
m: say "yay" if "hi" ~~ uc => "HI"
camelia rakudo-moar d8fa56: OUTPUT«yay␤»
01:26 rmgk left
timotimo m: say "yay" if "hi" ~~ :say 01:26
camelia rakudo-moar d8fa56: OUTPUT«hi␤yay␤»
timotimo simple as that
Sgeo timotimo, doc.perl6.org/type/Pair
I don't see an ACCEPTS listed, it would be helpful if it was 01:27
timotimo ah
yes, very true
01:27 rmgk joined
Sgeo Also that : isn't required? 01:27
m: say "yay" if "hi" ~~ say
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5===␤Argument to "say" seems to be malformed␤at /tmp/8IZkeuxgR_:1␤------> 3say "yay" if "hi" ~~ say7⏏5<EOL>␤Other potential difficulties:␤ Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an …»
timotimo it is
Sgeo Oh I missaw the uc one
I can imagine someone trying .e but that won't work except in a when because it's against the $_ not against the argument 01:28
timotimo "the argument"? 01:29
oh 01:30
you mean "foo.txt" ~~ .e
Sgeo yes
timotimo you'll get a warning, but you'll not likely immediately get what it tries to tell you
m: given "foo.txt" { when .e { say "hi" } }
camelia rakudo-moar d8fa56: OUTPUT«Method 'e' not found for invocant of class 'Str'␤ in block <unit> at /tmp/LtKKhM3dv9:1␤␤»
timotimo m: given "foo.txt" { when :e { say "hi" } } 01:31
camelia rakudo-moar d8fa56: OUTPUT«Method 'e' not found for invocant of class 'Str'␤ in block <unit> at /tmp/JPWu7VcWCv:1␤␤»
Sgeo THat when would work though because of the given
timotimo oh, also, the .e and stuff don't work on strings directly, you'll need to .IO that
Sgeo It's ~~ directly that could be an issue
timotimo m: given "foo.txt".IO { when :e { say "hi" } }
camelia ( no output )
timotimo m: given "foo.txt".IO { when .e { say "hi" } }
camelia ( no output )
timotimo er, huh
m: say "foo" ~~ True
camelia rakudo-moar d8fa56: OUTPUT«Potential difficulties:␤ Smartmatch against True always matches; if you mean to test the topic for truthiness, use :so or *.so or ?* instead␤ at /tmp/5ZNEMnKVB9:1␤ ------> 3say "foo" ~~7⏏5 True␤fooTrue␤»
timotimo m: say "foo" ~~ say "hi"
camelia rakudo-moar d8fa56: OUTPUT«hi␤True␤»
timotimo m: say "foo" ~~ (say "hi")
camelia rakudo-moar d8fa56: OUTPUT«hi␤True␤»
timotimo it only warns if the True or False is syntactic, i see.
Sgeo m: "foo.txt".IO ~~ .e 01:32
camelia ( no output )
01:32 HoppingMadMan left
Sgeo m: say "foo.txt".IO ~~ .e 01:32
camelia rakudo-moar d8fa56: OUTPUT«False␤»
Sgeo m: say "foo.txt".IO ~~ :e
camelia rakudo-moar d8fa56: OUTPUT«False␤»
timotimo oh, hold on
m: say "hello there" ~~ .say
camelia rakudo-moar d8fa56: OUTPUT«hello there␤True␤»
timotimo m: "hello there" ~~ .say
camelia rakudo-moar d8fa56: OUTPUT«hello there␤»
timotimo $_ is set on the RHS of ~~
Sgeo Ah
01:46 [particle]1 joined 01:48 gfldex left 01:50 Hor|zon joined 01:54 Hor|zon left 01:59 llfourn joined 02:04 llfourn left 02:13 yqt left
Sgeo "In the case of my $x, the lexpad entry for the variable $x is a pointer to an object of type Scalar, usually just called the container." 02:16
Is this still true? I don't see Scalar on the list of types
ShimmerFairy m: my $a = 42; say $a.WHAT; say $a.VAR.WHAT 02:17
camelia rakudo-moar d8fa56: OUTPUT«(Int)␤(Scalar)␤»
ShimmerFairy As far as I can tell, yes, and Scalar should probably be on a list of Perl 6 types :) 02:18
02:19 Zoffix left
Sgeo m: my $a; my $b := ($a, 0); $b[0] = 5; say $b; 02:22
camelia rakudo-moar d8fa56: OUTPUT«5 0␤»
Sgeo Can I imagine $ and @ to be namespaces where default assignment to a non-rebinded thing in that namespace behaves different? 02:23
Or what can illustrate the difference between $ and @? 02:24
ShimmerFairy m: my $a = [1,2]; my @b = [1,2]; my @c = 1,2; say $a.perl; say @b.perl; say @c.perl; # here's a difference 02:26
camelia rakudo-moar d8fa56: OUTPUT«[1, 2]␤[[1, 2]]<>␤[1, 2]<>␤»
Sgeo m: my $a = 0; $a[0]=5; say $a; 02:32
camelia rakudo-moar d8fa56: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/hKWAZrBNeh:1␤␤»
Sgeo Works with @
02:33 Guest44551 is now known as ponbiki 02:46 noganex joined 02:48 noganex_ left
zostay what's the difference between [] and []<>? 02:49
02:49 cognominal left
zostay ah... []<> flattens and [] does not? 02:50
02:51 kaare_ joined 03:00 davido__ left 03:01 davido__ joined
skids zostay: []<> is going away. 03:02
It was a pre-GLR experiment/workaround. 03:03
(As far as I know, at least)
Sgeo If I have a capture like \($a, $b), am I capturing the containers or values? Can I pass that to things expecting is rw? 03:05
Did I get Capture syntax right?
zostay syntax is correct
Sgeo ty 03:06
zostay i don't know the answer to your first question, though
PerlJam Sgeo: you can find out what Rakudo thinks with a little experimentation :) 03:09
Sgeo m: sub foo($a is rw, $b) { $a = 5; say $a; }; my $f = 10; my $x = \($f, 12); foo(|$x);
camelia rakudo-moar d8fa56: OUTPUT«5␤»
PerlJam A "say $f;" at the end could have told you the value/container answer. 03:11
Sgeo Huh, where? $f wouldn't be transformed. Maybe a say $x? 03:15
03:15 dayangkun joined
Sgeo m: sub foo($a is rw, $b) { $a = 5; say $a; }; my $f = 10; my $x = \($f, 12); foo(|$x); say $x; 03:15
camelia rakudo-moar d8fa56: OUTPUT«5␤\(5, 12)␤»
Sgeo m: sub foo($a is rw, $b) { $a = 5; say $a; }; my $f = 10; my $x = \($f, 12); foo(|$x); say $x.perl;
camelia rakudo-moar d8fa56: OUTPUT«5␤\(5, 12)␤»
Sgeo ??
skids m: sub foo($a is rw, $b) { $a = 5; say $a; }; my $f = 10; my $x = \($f, 12); foo(|$x); say $f; 03:20
camelia rakudo-moar d8fa56: OUTPUT«5␤5␤»
Sgeo say @bar.perl; # [7, 9]<> 03:23
If a doc says that, is it old?
Because if it is, I'll be sad because it's the clearest $ vs @ I've seen
doc.perl6.org/language/variables
skids It's current/recent but people did not like it. 03:24
Sgeo So something's changing? 03:25
:/
skids Yes we still have the GLR to go before release.
03:27 yeahnoob joined
Sgeo m: my $foo\ bar = 5; say $f\ oobar 03:29
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/p9fCJi8QWz␤Bogus postfix␤at /tmp/p9fCJi8QWz:1␤------> 3my $foo7⏏5\ bar = 5; say $f\ oobar␤ expecting any of:␤ infix␤ infix stopper␤ statement end␤ statement …»
Sgeo I was expecting that to work, I thought those are unspaces 03:30
Ah "Although we say that the unspace hides the whitespace from the parser, it does not hide whitespace from the lexer. As a result, unspace is not allowed within a token." 03:31
03:42 ChoHag joined
ChoHag In particular, a role may say 03:43
trusts ::?Class;
to allow self!attr() access to the role's $!attr variables with the class or from other roles composed into the class.
Surely self!foo() allows access to the *methods*, which may or may not be accessors. How are trustworthy attributes accessed? 03:44
(Also: 'from the class')
(Also also: Amongst pet examples, it discusses using proto to disambiguate conflicting methods and uses the example 'proto method shake {...}' - shaking pets is bad) 03:47
03:47 davido__ left 03:48 davido__ joined
ChoHag (From the top of S14) 03:49
Sgeo Is subset basically like contracts except with nicer syntax than in Racket? 03:50
How careful are they to maintain correct blame in higher-order function situations?
*{$x} -> %h { %h[$x] } 03:51
03:51 Hor|zon joined
Sgeo Is that correct or should it be 03:51
*{$x} -> %h { %h{$x} }
design.perl6.org/S02.html
03:56 Hor|zon left
skids That's a typo I'm pretty sure. 03:56
ShimmerFairy Yeah, using [] on a hash just does the normal "non-list is treated as one-item list" thing, which is certainly not what you want in that situation. 03:57
Sgeo m: (-> $x { say "foo"; })(1|2); (sub ($x) { say "bar"; })(1|2) 04:00
camelia rakudo-moar d8fa56: OUTPUT«foo␤bar␤bar␤»
Sgeo Also that mandatory space after sub for that second statement threw me for a loop, Javascript doesn't need that
04:00 llfourn joined
Sgeo So why is Mu default for the first and Any default for the second? 04:00
skids Sgeo: general and consistent rule (I think) is all "ident()" are calls. 04:04
Sgeo If you want to be able to match normal values and Junctures it's preferred to use Any|Juncture, in case there are future things added to Mu-but-not-Any?
04:05 llfourn left
Sgeo m: my $failure = fail "I fail."; say "No fail."; 04:07
camelia ( no output )
04:07 virtualsue joined
Sgeo Shouldn't it have either failed or succeeded? 04:07
dalek ecs: e2ee789 | skids++ | S02-bits.pod:
Fix a copy-paste that had left hash example with wrong brackets
skids Sgeo++ 04:08
ShimmerFairy m: say Nil ~~ Any; say Failure ~~ Any; 04:09
camelia rakudo-moar d8fa56: OUTPUT«True␤True␤»
skids Sgeo: No because you never sunk the failure.
ShimmerFairy Huh, S02 says this shouldn't be the case:
skids m: my $failure = fail "I fail."; $failure; say "No fail."; 04:10
camelia rakudo-moar d8fa56: OUTPUT«WARNINGS:␤Useless use of variable $failure in sink context (line 1)␤»
skids Oh I guess that gets caught these days.
Sgeo skids, if I never sunk the failure, why didn't the say go through?
ShimmerFairy > Any Perl 6 object (default routine parameter type, excludes Junction, Nil, Failure)
skids Hrm good question, it does with -e. 04:11
ShimmerFairy Oh, I think I got the smartmatches reversed, nvm 04:13
Sgeo Tieable role should be documented imo 04:14
skids Oh right. fail() actually tries to return unlike Failure.new. And right now there's a bug when it tries to do it at the top level.
Sgeo Also tying reminds me of Tcl 04:15
core.tcl.tk/tcllib/doc/trunk/embed...e/tie.html
A more complicated API around something Tcl itself provides in a simple manner, lemme see
skids m: sub a { my $failure = fail "I fail."; say "No fail."; }; a(); #...but... 04:16
camelia rakudo-moar d8fa56: OUTPUT«I fail.␤ in sub a at /tmp/p4nGRo7QVH:1␤ in block <unit> at /tmp/p4nGRo7QVH:1␤␤Actually thrown at:␤ in block <unit> at /tmp/p4nGRo7QVH:1␤␤»
skids m: sub a { my $failure = Failure.new("I fail."); say "No fail."; }; a(); #...but...
camelia rakudo-moar d8fa56: OUTPUT«No fail.␤»
Sgeo tcl.tk/man/tcl8.6/TclCmd/trace.htm#M14
skids m: sub a { my $failure = Failure.new("I fail."); $failure; say "No fail."; }; a();. 04:17
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/92XB3tXRkD␤Bogus statement␤at /tmp/92XB3tXRkD:1␤------> 3l."); $failure; say "No fail."; }; a();.7⏏5<EOL>␤ expecting any of:␤ dotty method or postfix␤»
Sgeo "trace add variable name ops commandPrefix
: Arrange for commandPrefix to be executed whenever variable name is accessed in one of the ways given by the list ops.
skids m: sub a { my $failure = Failure.new("I fail."); $failure; say "No fail."; }; a();
camelia rakudo-moar d8fa56: OUTPUT«No fail.␤»
skids hrm.
Sgeo Tcl has a few too many ways to pass code to be executed to something 04:18
As in, a subroutine might expect a script, or a command prefix, or (I forget what else)
Command prefix is the modern way, but Tk is old and expects scripts
ChoHag Haha! Expects scripts. Because expect is written in ... never mind 04:19
skids Something strange with camelia's context WRT failures I guess.
Sgeo ChoHag, heh, forgot about that, pun not intended
ChoHag Forgetting about expect is a net positive. 04:20
I would like to do the same.
Sgeo I never actually used it. 100% of my interest in Tcl is because I like learning about languages, although it did come in handy on the job because needed to add a thing to a load balancer that uses Tcl 04:21
s/it/expect/ 04:22
skids Yeah every once in a while I'm like "crap Im going to have to lear TCL" because it is popular on network gear, but a better solution than device-side scripts seems to always appear. 04:23
Thank goodness most of the next-gen stuff is moving to python, not that I know that either but... 04:24
ChoHag Python's easy.
Sgeo Tcl is actually... interesting IMO, in its model of metaprogramming
But I very strongly recommend learning Tcl via tutorial instead of example
skids Yeah but metaprogramming is the last thing you want happenning on routers :-) 04:25
ChoHag But (and probably because) it's not very expressive.
Sgeo Tcl is beautiful if learned properly, horrific trying to guess $ vs no $ and expr and all these other things if not learned properly
Although due to lack of garbage collection of mutable things, I tend to... be really really bad at not leaking gigs of memory 04:26
"That is, if $a interpolates, so do $^a, $*a, $=a, $?a, $.a, etc. It only depends on the $." 04:29
IIUC, $a does not interpolate? Not that the text is wrong if I'm understanding correctly, but maybe a better example could be used?
ShimmerFairy $a does interpolate in places that allow it. 04:30
Sgeo "It will not automatically dereference to its contents unless placed explicitly in some kind of dereferencing context." 04:32
What are dereferencing contexts 04:33
skids I think prety much everything but a binding context? 04:35
Sgeo lol context 04:36
04:36 BenGoldberg left
ShimmerFairy Sgeo: in the part you're reading (AFAICT), it's saying that $x will never expose itself as anything aside from a singular item _unless_ you put it in @($x) or such. 04:36
Sgeo m: my $a = (1,2,3); my @b = $a, 4, 5; say @b;
camelia rakudo-moar d8fa56: OUTPUT«1 2 3 4 5␤»
04:37 vendethiel joined
skids m: my $a = (1,2,3); my @b = $a, 4, 5; say @b.perl; 04:37
camelia rakudo-moar d8fa56: OUTPUT«[1, 2, 3; 4; 5]<>␤»
Sgeo Ah 04:40
ShimmerFairy The .gist method for lists tends to be lacking, in my opinion. 04:42
skids up too late again. Gotta stop dieseling and go to bed. 04:44
04:44 skids left 04:45 cognominal joined 04:46 nys left 04:47 [Sno] left
Sgeo Can a parcel be interpreted as a capture at runtime? THat is, a thing with a dynamically created Pair being used to mean named argument? 04:48
ChoHag Is this a bug or expected behaviour: 04:51
m: class A { multi submethod BUILD { say "...A" } }; class B is A { submethod BUILD { say "...B" } }; class C is A { }; say "B:"; B.new; say "C:"; C.new
camelia rakudo-moar d8fa56: OUTPUT«B:␤...A␤...B␤C:␤...A␤...A␤»
ChoHag vs.
m: class A { submethod BUILD { say "...A" } }; class B is A { submethod BUILD { say "...B" } }; class C is A { }; say "B:"; B.new; say "C:"; C.new
camelia rakudo-moar d8fa56: OUTPUT«B:␤...A␤...B␤C:␤...A␤»
ChoHag Seems like A's submethod is leaking into C when it's a multi. 04:53
04:53 llfourn joined 05:00 vendethiel left
Sgeo doc.perl6.org/type/Pair 05:10
Should probably have the :x($foo) syntax
ShimmerFairy it does, though. (see :key<value> near the top) 05:12
Sgeo That's for strings though, iiuc 05:13
05:13 virtualsue left
ShimmerFairy yeah, but it's just sugar for :key('value') 05:16
Sgeo Might not be obvious to a newbie reading the docs? 05:17
ShimmerFairy Fair enough, though there might be some sort of "general overview" page describing it already. (Certainly wouldn't hurt, I don't think.)
tony-o ChoHag: that looks like a bug 05:19
05:22 khw left
Sgeo m: grep Failure.new("FAIL"), 0, 1 05:23
camelia ( no output )
jdv79 tony-o: why did you create 2 HTTP::Server dists?
tony-o jdv79: what do you mean? 05:24
one is a role..
jdv79 Async and Threaded seem odd name choices as rakudo's async uses threads.
tony-o one uses an ::Async socket, the other uses blocking sockets in threads
jdv79 they both look like classess to me 05:25
ah
tony-o i thought you meant HTTP::Server - not those two things
jdv79 do you see both being used long term or just for experimentting? 05:26
tony-o the ::Threaded one will go away, it was temporary while the IO::Socket::Async problems were hammered out 05:27
they're interchangable, though, both abide by 'HTTP::Server' 05:28
Sgeo Does it make sense to have both async and sync? If some library uses sync it becomes harder to do async stuff with that library. That's, like, a good thing (of few) that Node.JS does, force the entire ecosystem async with cultural and technical pressure 05:30
ChoHag Then don't use that library if you want async code. 05:31
jdv79 easier said than done 05:32
sometimes there is no alternative
ChoHag Do you not have a text editor?
(Yes yes. I know. I'm joking ... mostly) 05:34
tony-o they both use threads, they just use them differently 05:35
ChoHag If I have a class B is class A, and class C is class A, can I (at run time) create an instance of class B is class C? 05:40
05:45 rurban joined
ChoHag NB. Not "should I", "can I". 05:45
05:46 laouji left, laouji joined 05:48 diana_olhovik_ joined
tony-o do you care if A is a role ? 05:51
05:52 Hor|zon joined
tony-o m: class A { has $.x = 5; }; class B is A { }; role C is A { }; my B $r .=new; $r does C; $r.x.say; 05:53
camelia rakudo-moar d8fa56: OUTPUT«5␤»
tony-o m: class A { has $.x = 5; }; class B is A { }; role C is A { has $.y = 50; }; my B $r .=new; $r does C; $r.y.say;
camelia rakudo-moar d8fa56: OUTPUT«50␤»
tony-o i'm too tired for this
ChoHag Yes because this is a thought experiment and I already know I can get a similar effect with roles.
This is just sticking fingers in the plug socket to see what happens, not anything practical. 05:54
If you want something practical and easy before you finally succumb to sleep, tell me how to distinguish at runtime between an instance and a type.
ugexe .DEFINITE? 05:55
05:56 Hor|zon left
ChoHag Ah! "There is one more macro:", just below the visible page. 05:57
05:58 FROGGS_ left 05:59 domidumont joined 06:00 rurban left
moritz \o 06:03
blogs.perl.org/users/sterling_hanen...rface.html
06:08 [Sno] joined 06:10 domidumont left 06:11 domidumont joined
jdv79 zostay++ 06:14
06:14 domidumont left 06:15 domidumont joined 06:17 telex left 06:18 telex joined 06:22 domidumont left 06:35 zacts joined 06:45 krunen joined 06:48 FROGGS joined 06:53 Hor|zon joined 06:58 Hor|zon left
Sgeo m: my @a = 1,2,3; say map *+1, @a 06:58
camelia rakudo-moar d8fa56: OUTPUT«2 3 4␤»
Sgeo But map seems to use vararg, so how does it take @a like that? 06:59
Does @a become multiple arguments in that context or something?
s/vararg/slurpy/ # Use Perl6 terminology
ChoHag Can anything other than the first argument be the routine to apply? 07:00
In which case all subsequent arguments are values to map over, and @a flattens. 07:01
m: my @a = [1,2,3]; say map *+1, @a; 07:02
camelia rakudo-moar d8fa56: OUTPUT«4␤»
moritz note that [] does not flatten in slurpies
ChoHag m: my @a = [1,1,1]; say map *+1, @a; 07:03
camelia rakudo-moar d8fa56: OUTPUT«4␤»
ChoHag I know.
Hooray! I had (whether given or not) an answer rather than a question!
<-- Progressing.
07:10 darutoko joined 07:12 azawawi joined
azawawi hi 07:12
moritz \o azawawi
07:18 RabidGravy joined
Sgeo m: say [1,2,3].WHAT 07:21
camelia rakudo-moar d8fa56: OUTPUT«(Array)␤»
Sgeo Do I need to know Perl 5 to make sense of all this context stuff?
ChoHag It might help if you didn't. 07:23
07:23 brrt joined
dalek c: ef6ea00 | moritz++ | / (2 files):
Document X::Channel::SendOnClosed
 07:24
c: 3446cd5 | moritz++ | lib/Type/ (2 files):
Document X::Channel::ReceiveOnClosed
07:33 pdcawley joined
RabidGravy morning! 07:38
jdv79 is it possible to make an obj look like a scalar and a list? 07:40
brrt yes, why not?
jdv79 how? 07:41
brrt m: my $a = [1,2,3]; say $a;
camelia rakudo-moar d8fa56: OUTPUT«1 2 3␤»
jdv79 cause i can't seem to do it.
07:41 virtualsue joined
brrt then you'll have to specify your question a bit more :-) 07:41
jdv79 no, i mean $foo = "asdf" would set $foo's scalar member and $foo.[1] would set into $foo's list member 07:42
brrt hmmm 07:50
i think you can fake that
but why would you
jdv79 i'm trying an overly clever interface for a path, either operate on the whole in str form or operate on the seperate parts 07:51
surely p6 is up to the task;)
llfourn do you have a MyPath obj? If so I think you can just define a postcircumfix for MyPath... 07:52
jdv79 that gets me list ops, sure
llfourn though im not sure how assignment like MyPath $path = '/foo/bar' would work
jdv79 but how about $foo = '/bar/baz'?
yeah
RabidGravy overr-ride assignment ;-) 07:53
llfourn I think in future you can do MyPath() $path to coerce it
jdv79 i think that is not allowed
07:54 Hor|zon joined
RabidGravy it's just make it a string sub-class with some list-like features 07:54
then the assignment would Just Work™ 07:55
jdv79 that's what i'm asking about
i don't know how to do it if its possible even
RabidGravy right, yiour question about assignment confused me
llfourn m: class Path is Str { }; my Path $p = '/foo/bar'; 07:56
camelia rakudo-moar d8fa56: OUTPUT«Type check failed in assignment to '$p'; expected 'Path' but got 'Str'␤ in block <unit> at /tmp/E1SZkXxAvL:1␤␤»
llfourn How do you make subclass str assignment work to string? 07:57
dalek p: 3fec6e0 | hoelzro++ | src/vm/jvm/ (2 files):
Create mapping and implementation for force_gc on JVM
p: a317d34 | hoelzro++ | src/vm/jvm/stage0/ (10 files):
Re-bootstrap JVM files for force_gc
p: 426ca43 | hoelzro++ | src/vm/jvm/HLL/Backend.nqp:
Add force_gc implementation for JVM backend
RabidGravy I lied
p: e99940c | FROGGS++ | src/vm/jvm/ (13 files):
Merge pull request #240 from perl6/jvm-force-gc

JVM force_gc
07:59 Hor|zon left
Sgeo How do you do a named argument slurpy that collects names that are used multiple times? 08:00
:@x collects multiple xs, but *% is used for named slurpy 08:01
RabidGravy jdv79, would applying a role to the attribute work for you?
jdv79 example =?
RabidGravy Speo 08:02
jdv79 maybe in a lame way i can "is Proxy" to get hold of assignment ^H
Sgeo If I call
foo :x(5) :x(6)
RabidGravy Speo, er you'd have to create the hash from a bunch of pairs
Sgeo ANd foo wants all named arguments
llfourn Sgeo: I think named args have to be unique, you want a list of pairs
Sgeo m: sub multix(:@x) {say @x; }; multix :x<hi> :x<bye>; 08:03
camelia rakudo-moar d8fa56: OUTPUT«Type check failed in binding @x; expected 'Positional' but got 'Str'␤ in sub multix at /tmp/cqCf3Nm6b8:1␤ in block <unit> at /tmp/cqCf3Nm6b8:1␤␤»
llfourn m: sub multix(:@x) { say @x }; multix [ :x<hi>, :x<bye> ]; 08:04
camelia rakudo-moar d8fa56: OUTPUT«Too many positionals passed; expected 0 arguments but got 1␤ in sub multix at /tmp/niUx8P02Pp:1␤ in block <unit> at /tmp/niUx8P02Pp:1␤␤»
llfourn m: sub multix(:@x) { say @x }; multix :x([ :x<hi>, :x<bye> ]);
camelia rakudo-moar d8fa56: OUTPUT«x => hi x => bye␤»
Sgeo Look for this: sub fun (Int :@x) { ... } 08:05
design.perl6.org/S06.html
llfourn Sgeo: You're right, looks like not yet implemented then.. 08:06
Sgeo So... how to slurpy?
And also I assume this will be fixed by December?
llfourn We can only hope :)
m: sub foo(*@pairs) { say @pairs }; foo x => 1, x => 2; 08:07
camelia rakudo-moar d8fa56: OUTPUT«Unexpected named parameter 'x' passed␤ in sub foo at /tmp/fdsfnW1ZuW:1␤ in block <unit> at /tmp/fdsfnW1ZuW:1␤␤»
llfourn m: sub foo(@pairs) { say @pairs }; foo [x => 1, x => 2]; 08:08
camelia rakudo-moar d8fa56: OUTPUT«x => 1 x => 2␤»
Sgeo m: sub foo(*%pairs) { say %pairs }; foo x => 1, x => 2;
camelia rakudo-moar d8fa56: OUTPUT«x => 2␤»
llfourn that's only way I can think of
yeah % will always uniqueify
Sgeo Maybe by requesting the capture? 08:11
08:11 vytas joined
Sgeo m: sub foo(|cap) { say cap }; foo x => 1, x => 2; 08:11
camelia rakudo-moar d8fa56: OUTPUT«\(:x(2))␤»
Sgeo m: sub foo(|$cap) { say $cap }; foo x => 1, x => 2; 08:12
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/EO8X5hd6T3␤Obsolete use of | or \ with sigil on param $cap␤at /tmp/EO8X5hd6T3:1␤------> 3sub foo(|$cap7⏏5) { say $cap }; foo x => 1, x => 2;␤ expecting any of:␤ shape declaration␤»
RabidGravy Speo, 08:13
m: sub f( *@x) {say @x}; f((x => 1), (x => 2))
camelia rakudo-moar d8fa56: OUTPUT«x => 1 x => 2␤»
RabidGravy you can also de-nameify them by making the keys variavle
08:14 vytas left
Sgeo Please stop calling me Speo, it's Sgeo and Speo does not ping me 08:14
RabidGravy please type in a bigger font then ;-p 08:15
llfourn sub f(*@x) {say @x}; f (x => 1, x => 2) 08:16
RabidGravy m: sub f( *@x) {say @x};my $x ="x"; f($x => 1, $x => 2)
llfourn m: sub f(*@x) {say @x}; f (x => 1, x => 2)
camelia rakudo-moar d8fa56: OUTPUT«x => 1 x => 2␤»
llfourn seems like they are the same :)
the space before the openning ( actually matters in p6 08:17
08:19 virtualsue left 08:21 sqirrel joined
jnthn morning, #moarvm 08:21
yoleaux 2 Aug 2015 12:09Z <nine> jnthn: your commit 56ae33ea0a5ac3d54a5fef2aa15eab34f9e9594b broke &?CALLER::ROUTINE: role Marker { }; sub foo { say &?CALLER::ROUTINE ~~ Marker; }; my $test = sub { foo(); }; $test(); $test does Marker; $test();
2 Aug 2015 14:32Z <nine> jnthn: I can workaround missing &?CALLER::ROUTINE with nqp::getcodeobj(nqp::ctxcode(nqp::ctxcaller(nqp::ctx)))
nine Good morning, jnthn!
nwc10 jnthn: but this is #perl6! 08:22
jnthn oops!
good morning, #perl6 :)
nine: &?ROUTINE access through CALLER only worked before by accident; note that none of &?BLOCK/$?FILE/$?LINE are available that way either.
nine: I don't intend to put it back; ratehr, that functionality should be exposed through callframe. 08:23
*rather
nine jnthn: ok. Unfortunately callframe doesn't give any access to the CALLER ROUTINE either. But at least nqp does. 08:24
jnthn nine: Yes, I know callframe doesn't at the momnet; I was saying that it should get a method that calls nqp::getcodeobj(nqp::ctxcode, ...) to let you get hold of it. 08:25
nine Shall we call it .routine? 08:26
jnthn No, 'cus not everything is a routine
Hm, I see that CallFrame lacks .caller and .outer methods to navigate around too 08:28
.code would perhaps be general enough
nine S06 says: "(For instance, caller.hints('&?ROUTINE') will give you the caller's routine object."
azawawi hmmm what's wrong with this code: $method = $method.subst(/ (<[A..Z]>) /, {$0.lc}, :g);
nine jnthn: I'd be fine with .code. I can have a shot at implementing the mentioned methods over the next couple of days. 08:29
08:32 Hor|zon joined
jnthn nine: Wow, that's...ugly :) Please lets just make it .code, to match .line and .file. ;) 08:38
nine: Also, please do feel free to take that one; I need to use what brane I have for GLR. 08:39
DrForr .file, .line, .column ? - Though I imagine there *are* some people out there that read their code in proportional fonts... 08:40
jnthn DrForr: We can only present information we actually have. :)
(We don't store column in the bytecode annotations) 08:41
08:41 azawawi left
DrForr Ah. 08:41
nine jnthn: will do :)
08:42 rindolf joined
jnthn From backlog: you can't collect all arguments with the same name, later ones override earlier ones. I know S06 speculates on something with the @ sigil in the receiving method, but it's actually pretty much impossible to implement (to see why, think about how it interacts with multi-dispatch, or Capture literals) 08:47
08:50 ChoHag left
dalek c: 21fdb8b | moritz++ | lib/Type/Channel.pod:
Channel: document what happens when calling .close while a .receive is blocking
 08:55
08:57 spider-mario joined 09:05 vendethiel joined 09:08 sqirrel left 09:10 vytas joined, virtualsue joined 09:11 Hor|zon left 09:20 vytas left 09:24 vytas joined
moritz zostay: ping 09:31
zostay: regarding P6SGI: on the one hand I love the ability to return a Promise or a Channel; OTOH it makes writing middlewares much more complicated, because a middleware must implement all the calling conventions, no? 09:32
I wonder if there's a simpler solution 09:33
jnthn I read the doc and need to ponder it more, but I suspect a Supply might allow better throughput for the streaming stuff
moritz that was another thing I considered 09:34
09:34 azawawi joined
moritz but supplies are tricky in this situation, no? 09:35
jnthn How so?
moritz because the caller must be able to tap before any events are emitted
or is that only a problem for on-demand supplies?
jnthn Oh, I was more thinking the API would return a Supply that is tapped in order to process the request. 09:36
Which will be much neater with the new "supply" keyword
Looking at the examples: 09:37
sub app(%env) {
start {
...
Is how you write something to process a request off in the thread pool
09:37 espadrine joined
jnthn So you can use await, presumably 09:37
Though I guess maybe to spread computation
But with what I'm suggesting the streaming case becomes more like: 09:38
sub app(%env) {
supply {
...
Which middleware - if it's not already on a thread pool thread - can just ".tap-on(ThreadPoolScheduler).tap(&send-output-to-client) 09:39
"
This gives much better thread affinity for the data in question 09:40
And so CPU cache locality
leont Yeah, supplies would make more sense to me too, but that's only after having worked with them for some time. To some extend it's an ackquired taste I guess.
09:41 yeahnoob left
jnthn Yeah, it's a producer/consumer threading model vs. a reactive one. 09:42
Generally, though, good performance depends on keeping inter-thread communication down.
And so far as I can see, that's only going to become more and more the case. 09:43
So any place that lacks natural temporal decoupling to render the cache effects of the inter-thread communication unimportant will suffer when you make unneeded movement of a flow between threads. 09:44
And that's the case here really: you just produced a bunch of data and you want to toss it into a socket (maybe after some further processing).
09:45 RabidGravy left
jnthn The Channel approach means you shove inter-thread communication there, and if the next layer down wants to actually process the data then the thread doing so will have to grab all of the data into its cache, while with the Supply one you're still on the same thread. 09:46
If anyone with a blogs.perl.org account wants to post a link to this, feel free 09:47
jnthn gets back to GLRing :)
brrt S32-io/IO-Socket-Async.rakudo.moar hangs, but you're probably aware of that 09:48
jnthn brrt: I wasn't; haven't seen it do that in a long, long time... :( 09:49
brrt oh :-(
well, you probably know i'm on linux 09:50
and maybe it helps if i say that i use unoptimized builds?
jnthn Maybe
Does it happen repeatably?
brrt quite :-) 09:54
09:58 g4 joined, g4 left, g4 joined
azawawi tony-o: ping 10:03
10:03 darutoko- joined
dalek c: 0441841 | smls++ | lib/Language/subscript (2 files):
rename page 'subscripting' -> 'subscripts'
 10:04
10:06 aindilis` joined 10:07 TimToady_ joined 10:08 araujo left, darutoko left, silug_ left, TimToady left, ponbiki left, rindolf left, ponbiki_ joined, aindilis left, rindolf joined, smls joined 10:11 araujo joined 10:16 silug_ joined 10:17 brrt left 10:22 RabidGravy joined 10:26 pierrot left 10:30 pecastro joined 10:31 leont left 10:35 renormalist joined 10:36 sqirrel joined
dalek c: fb90586 | moritz++ | lib/Language/subscripts.pod:
Fix typo, LLFourn++
 10:42
smls oops :P 10:45
nine Posted an irc.perl6.org link to the P6SGI blog post 10:51
10:54 pierrot joined
dalek c: 32aa3fd | smls++ | lib/Language/subscripts.pod:
subscripts: fix :delete section example; ronaldxs++
 10:57
tony-o azawawi: whats up? im getting ready to leave for the airport? 11:05
s/\?$//
azawawi tony-o: any plans to integrate Bailador with HTTP::Server::* ?
11:07 sqirrel left
tony-o what do you mean? 11:07
azawawi tony-o: Bailador to use HTTP::Server::Threaded as a backend instead of HTTP::Easy::PSGI 11:08
tony-o if youd like to and you build it around the http::server role - http::server::async is almost ready 11:09
azawawi Bailador::Async ?
or Bailador::Threaded... 11:10
tony-o i didnt have anything i intended to write right away or anything
azawawi ok on it :)
thanks for the reply
tony-o after async, i have like 10 other modules that are probably broken :/) 11:11
azawawi like? 11:12
11:21 pierrot left
nine The modern way would be a PSGI (or P6SGI) backend for Bailador and HTTP::Server::* based PSGI implementations. 11:22
tadzik well, internally that's how it works :)
nine shuts up then
tadzik bailador is only using http-easy because it was the only thing that worked anyway
or so I remember :) 11:23
yeah, Bailador::dispatch_psgi seems to be the psgi app
github.com/tadzik/Bailador/blob/ma...or.pm#L108 11:24
azawawi tadzik: i see 11:26
nine: thanks :)
tony-o psgi seems dated 11:27
RabidGravy well it's six or so years old 11:33
11:37 TEttinger left 11:39 Zoffix joined 11:42 gfldex joined 11:43 brrt joined 12:10 zakharyas joined
itz much of the perl5 world probably still uses CGI TBH 12:11
Zoffix Probably. Legacy stuff and what not. 12:13
virtualsue a lot of the world is legacy
jnthn Legacy = the code made it into production and is doing useful stuff. Becoming legacy is success. :) 12:15
Zoffix Legacy = the code that is too expensive to upgrade. :) 12:16
jnthn We might often wish folks had left us a better legacy, alas... :)
brrt huh, i recently read 'technical depth' for 'technical debt' 12:18
and i thought, that's not even wrong 12:19
12:21 colomon joined
itz the code you wrote might be "legacy" but mine is "heritage" ;) 12:22
12:22 brrt left
Zoffix lol 12:23
12:24 kkielak joined 12:25 oxide94 joined 12:26 oxide94 left 12:31 leont joined
jdv79 self.bless returns an instance? 12:31
nine yes 12:32
m: class Foo { method new() { return self.bless } }; Foo.new.perl.say 12:33
camelia rakudo-moar d8fa56: OUTPUT«Foo.new␤»
RabidGravy m: enum G (Boo); # somewhat less that fabulous error message 12:35
camelia rakudo-moar d8fa56: OUTPUT«===SORRY!===␤Cannot invoke this object (REPR: Null, cs = 0)␤»
jdv79 i was getting : Cannot look up attributes in a type object
nine jdv79: with what code exactly? 12:36
12:36 TimToady_ left 12:37 lucasb joined 12:38 TimToady joined, kkielak left 12:39 dayangkun left
jdv79 now i'm getting a different error 12:41
gist.github.com/anonymous/4d6c3cff038c4b3131bd
trying to access a private array attr in new
RabidGravy you probably want to pass the a => 1,2,3 to bless 12:43
jdv79 the docs are lacking on this sort of thing ^H
nine The whole process is somewhat less magical than one would expect. bless returns an instance, but it doesn't magically replace self in your class method. self still is your type object
jdv79 ok, but i used $obj not self
nine If you need to access the new object's privates, you have to do so in an object method like BUILD
jdv79 why not in new? 12:45
RabidGravy when you create the object using bless, you are getting a new object, you are not in *its* new so the attribute is private to you 12:46
12:46 cognominal left
jdv79 seems unusual 12:47
and i didn't notice that in the docs
nine jdv79: it's really just a consequence of the construction process being as little magical as possible. The advantage is, that it's very customizable.
12:48 cognominal joined
RabidGravy just added something to that gist 12:49
jdv79 so to set a priv attr at construction one needs to pass to bless and defined a BUILD to catch it? 12:50
RabidGravy dunno why it shagged up the formatting thoug
well that's one way of doing it yes 12:51
jdv79 ok 12:52
jdv79 can't wait for comprehensive docs
i thought i had more patient to dig but its like every little thing
gets to be too much ^H:( 12:53
*patience
thanks
RabidGravy this is explained in the docs.perl6.org/language/classtut#Constructors
jdv79 ok. i see it now. 12:56
i was reading the spec and then hit that doc and missed taht bit 12:57
RabidGravy :) 12:59
13:02 spider-mario left 13:03 zakharyas left
dalek c: c470d2c | smls++ | lib/Language/subscripts.pod:
subscripts: Fill in the "Basics" section
 13:04
smls jnthn, lizmat: Since you seem to be the experts on that topic, it would be nice if one of you could take a look at doc.perl6.org/language/subscripts#Custom_types (everything from that heading to the end of the page) to check if what I wrote is correct and if the example code looks sane. 13:09
13:14 azawawi left
PerlJam smls++ nice example 13:15
timotimo oi 13:16
er
o/
do we really want the heading as well as the first line of tables rendered in yellow?
rather than having headings in a different color, so that the colors stay somewhat alternating? 13:17
or perhaps switch up even and odd in tables?
it's probably important to make the style work for tables with and without headers ...
PerlJam timotimo: I'm fine with the even/odd colors, but the headings need a different color IMHO 13:18
smls PerlJam: My intention for that example was to demonstrate all 3 different "levels" of providing AT-KEY: 1) simple delegation 2) explicit method that simply forwards another object's AT-KEY 3) method that returns a custom container
timotimo: I'm not really happy with the table style either. It's somewhere on my todo list.
timotimo good to know, thanks! 13:19
PerlJam smls: one random thing that occurs to be because of your doc is that .elems seems out of place. Or it makes me think that there should be an .ELEMS that .elems is automatically aliased to. 13:22
13:23 khw joined
smls Well, in general you'd also implement .list .keys .values .iterator etc 13:23
.elems just happens to be used by subscripts in addition to users calling it directly 13:24
13:24 awwaiid joined
smls but that's not so different from .list/.iterator being called by `for` or `map`. 13:24
PerlJam true 13:25
I guess my nice, neat mental box that says "uppercase may be called for you by Perl" doesn't want it to be any other way :) 13:26
smls everything is better than Python's ghastly __foo__ naming scheme for special/interpreter-called methods ;) 13:27
PerlJam I dunno ... those are certainly unambiguous :) 13:30
dalek ast: 86cfb03 | (Stefan Seifert)++ | S06-advanced/callframe.t:
CallFrame.code should now return the code object

As per jnthn++'s suggestion:
  irclog.perlgeek.de/perl6/2015-08-03#i_10994354
kudo/nom: a38b59c | (Stefan Seifert)++ | src/core/CallFrame.pm:
Support CallFrame.code to return the code object

As per jnthn++'s suggestion:
  irclog.perlgeek.de/perl6/2015-08-03#i_10994354
13:37 skids joined 13:47 brisby joined 13:50 synbot6 joined 13:51 synbot6 left 13:52 synbot6 joined, synbot6 left 13:53 laouji left
lucasb evalbot control version 14:00
camelia This is evalbot revision eb33872
lucasb idk if "evalbot control pull" is enough to update the evalbot
it seems there is a newer revision that makes github gist api work again 14:01
sorry! I issued 'evalbot control restart'. I thought it would come back automatically 14:06
I thought it was running inside a 'while true; do ...; done' loop 14:07
jnthn lucasb: It...usually does come back, from memory. 14:08
14:08 synbot6 joined, brisby left
RabidGravy boom 14:09
timotimo it ought to, yeah, i think so 14:15
nine: can you have a quick look-see?
lucasb I'm very very sorry! My intent was just update it. Someone with a hack account will have to take a look, I guess.
FROGGS it is not not running on hack AFAIK 14:16
timotimo correct
lucasb oh
timotimo it runs on nine's box
it's no problem, lucasb, we'll be back to working order in no time 14:17
RabidGravy and in the interim we can talk among ourselves ;-) 14:26
nine Can't locate JSON.pm in @INC (you may need to install the JSON module) BEGIN failed--compilation aborted at lib/EvalbotExecuter.pm line 79.. 14:31
timotimo huh, that's interesting 14:33
lucasb oh, right, Mouq++ added the JSON dependency when he fixed the github gist api
timotimo would it have to be JSON::PP or JSON::XS?
14:33 brrt joined
RabidGravy don't they have the same interface? 14:35
FROGGS I think so
nine JSON will use JSON::XS if installed
Is this something camelia should be able to do or should I try to install JSON into camelia's perlbrew perl? 14:36
FROGGS I don't think camelia can install Perl 5 modules
14:37 camelia joined, FROGGS left
nine m: say "I'm back :)" 14:37
RabidGravy got the arse now 14:38
14:38 ChanServ sets mode: +v camelia
camelia rakudo-moar a38b59: OUTPUT«I'm back :)␤» 14:38
lucasb thank you, nine++; and sorry for the trouble :) 14:39
evalbot control version
camelia This is evalbot revision 2267d41
lucasb m: gist.github.com/anonymous/d2efba968f2a8be140a2
camelia rakudo-moar a38b59: OUTPUT«hi␤»
lucasb works fine now :) 14:40
timotimo very good
RabidGravy fix it real good 14:42
hoelzro good morning, #perl6! 14:49
nine And a very good morning to you, hoelzro!
hoelzro o/ nine! 14:50
14:51 nys joined
b2gills Chrome crashed so I think it's time for $ apt-get upgrade && reboot 14:52
14:52 b2gills left
RabidGravy chrome has been really, really bad since the last update 14:53
geekosaur has not had problems with chrome since disabling gpu support 14:54
brrt has actually been using firefox for a while now
geekosaur somehow that makes it suck both cpu and memory without limit
brrt oh, and i can't for the life of me get nouveau to do something interesting
geekosaur every so often I fire up an updated firefox, promptly get reminded why I detest it, and go back to chrome 14:55
jnthn Funny, chrome crashed here today too...after running for a month solid.
Remembered my tabs across two windows <3
14:57 b2gills joined
brrt hmm, i actually hard shutdown my laptop usually 14:57
geekosaur that said, I update chrome as often as updates are released, so it's unusual (at least of late, "thanks" Hacking Team) for Chrome to reach a week of uptime ;) 14:58
b2gills My computer has been trying to get me to update Chrome, and the kernel for more than a week 15:00
15:02 [ptc] left
zostay jnthn, moritz: i'm not opposed to using a Supply instead 15:06
as for whether or not using a Promise/Channel/Supply will cause trouble for the middleware, that will happen anyway
if you want to write complete middleware in PSGI, you have to handle all the callback cases, generally done by using a Plack helper 15:07
completely functional middleware must implement everything for both the server and app sides of the spec 15:08
b2gills I think there is a better interface "hiding" in Perl 6 that no-one has "found" yet.
15:09 [ptc] joined, burnersk joined
moritz b2gills: maybe one could come up with a middleware design where dispatching to the application is just a callsame() came 15:09
*call 15:10
zostay i hadn't suggested it in the spec, but i don't see why middleware can't be applied by just calling .wrap() on the app routine 15:14
moritz one could also imagine an API built around returning promises (even in the synchronous cases) 15:15
zostay: well, the problem isn't calling; it's dealing with the return value
15:15 brrt left
moritz zostay: imagine a middleware that always wants to add a header ("Cache-Control", for example) 15:15
ugexe i use the promise method for synchronous stuff quite a bit
moritz zostay: so it has to do check the type of the return value, and do one thing if it's Positional, and another if it's a Promise, and a third if it's a Channel 15:16
zostay right... the only way around that is to require every app implement a promise that returns an iterable that returns a channel/supply for content 15:17
afaict
15:17 telex left
zostay the point is to make app writing easy for the app authors, and make server and middleware writing possible 15:17
if you use the channel or the promise interface you are making a statement that something about your performance is different than a typical app 15:18
15:18 telex joined
zostay that's the case if you return a callback in PSGI 15:18
moritz well, there's an alternative 15:19
one could pass around proper objects insteads of hashes / arrays
and those have event callbacks 15:20
so a middleware could register a on-header-available callback that adds a header, for example
zostay is there a generic type in P6 that does that? 15:21
moritz (I know this is quite a departure from the current API)
zostay: no; would need be a custom type
another option would be to supply such a type as a utilty/wrapper around the data structure
zostay then i demur and say you've described a high level framework, not P6SGI which is lowest-common-denominator 15:22
the reference implementation can provide helpers, but the standard itself must be low level
PSGI/Plack have demonstrated success in this
moritz agreed, but it's a small low-level API 15:23
the larger it becomes, the more painful its low-levelness becomes
15:26 diana_olhovik_ left
zostay i need to finish the reference implementation and then we can try things out better 15:34
nine In any case I'm happy to see things moving on this front :) 15:37
15:47 cognominal left 16:00 Begi joined
Begi Hey #perl6 ! How can I get the total number of lines in a file ? 16:01
Is there like perl 5 a $. var ? 16:02
PerlJam Begi: $*IN.ins (or some other filehandle)
m: END { say $*IN.ins } # curious 16:04
camelia rakudo-moar a38b59: OUTPUT«0␤»
PerlJam figured
m: END { say $*ARGFILES.ins } # how about this?
camelia rakudo-moar a38b59: OUTPUT«(Any)␤»
PerlJam that was unexpected 16:05
(I would have expected 0 alos)
16:20 spider-mario joined
timotimo $*ARGFILES isn't quite perfectly smart yet 16:23
16:40 rangerprice joined 16:48 yogan joined
Begi Ok thanks. And what's the easiest way to select the last 5 lines of a file ? 16:53
TimToady m: my @lines = lines(); say @lines[*-5 .. *] 16:54
camelia rakudo-moar a38b59: OUTPUT« Slán! Slán go fóill, a Dhún na nGall, a chontae shéimh gan smál, Is do d’fheara breáth’ in am an ghá nár úmhlaigh riamh roimh Ghall; Tá áit i mo chroí do gach fear is gach mnaoi is gach páiste beag agus mór Atá beo go buan gan bhuairt …»
Begi Ok thanks. Eh... Is camelia ok ? 16:57
moritz m: say 'yes, sir' 16:59
camelia rakudo-moar a38b59: OUTPUT«yes, sir␤»
Begi :P 17:00
17:00 DarthGandalf joined 17:01 colomon left
TimToady wonders if the Prince of Wales knows Welsh... 17:05
llfourn fairly sure he doesn't 17:06
itz he spoke some Welsh at his investiture 17:10
17:12 domidumont joined, colomon joined
RabidGravy he did indeed, but we're old enough to remember that 17:16
17:17 espadrine left
RabidGravy it of course could have been garbage as I for one don't know Cymraeg at all 17:19
horrifyingly I know more Catalan 17:20
vendethiel my "Programming Perl" books (page 1008, under the "use charnames;" section): `"AMELIA" => "DROMEDARY CAMEL"`. Did camelia change name? :P
dalek rl6-roast-data: b8b5cbe | coke++ | / (9 files):
today (automated commit)
 17:21
17:23 domidumont left
lucasb .u camel 17:24
yoleaux U+1F42A DROMEDARY CAMEL [So] (🐪)
U+1F42B BACTRIAN CAMEL [So] (🐫)
[Coke] .u moose
yoleaux U+1509 CANADIAN SYLLABICS MOOSE-CREE SK [Lo] (ᔉ)
lucasb No butterfly in unicode :( 17:26
17:27 mohij joined
[Coke] .u butt 17:30
yoleaux U+1F518 RADIO BUTTON [So] (🔘)
U+1F532 BLACK SQUARE BUTTON [So] (🔲)
U+1F533 WHITE SQUARE BUTTON [So] (🔳)
17:34 dha joined 17:40 khw left 17:45 lizmat joined 17:47 virtualsue left 17:53 Begi left 17:57 llfourn left
vendethiel are the "less" directives actually considered in P5? 17:57
17:58 aristotle joined 18:00 llfourn joined
aristotle zostay: I think you are being too all-or-nothing when you say if you use the channel or the promise interface you are making a statement that something about your performance is different than a typical app 18:00
18:01 FROGGS joined
aristotle zostay: I don’t think there’s any point in Perl 6 for distinguishing the deferred/immediate cases. while there may be a case for distinguishing the streaming case 18:02
lizmat good *, #perl6! 18:03
PSA: working on the P6W at the moment...
aristotle zostay: you don’t want to have to use the PSGI deferred style in Perl 5 if you don’t need to, because it’s so damn inconvenient, because the language doesn’t give you any better option
zostay: but in Perl 6 it’s just putting `start { }` around your code. that’s nothing
lizmat if anybody thinks of something that should be mentioned, please let me know! 18:04
18:05 llfourn left
aristotle zostay: the real inconvenience when writing middlewares is the deferred vs immediate case difference. the streaming vs immediate body difference would be simple to handle 18:06
18:06 zakharyas joined
[Coke] the upcoming swiss perl workshop, maybe? 18:06
(for liz)
lizmat [Coke]: already in there, but [Coke]++ for mentioning :-)
aristotle zostay: many middlewares (maybe most) don’t need to look at the response body at all, so would be unaffected by the body having two possible forms. but many middlewares need to modify the response headers, so they do have to handle the deferred vs immediate distinction 18:13
18:15 diana_olhovik joined, ChoHag joined
aristotle zostay: and I don’t see the value of making that distinction in Perl 6, since it doesn’t cost apps anything appreciable to uniformize the response (and that’s the only place where that might incur any costs. every other part of the system becomes simpler for having fewer cases to deal with) 18:16
moritz, leont: ping ^
18:19 pierrot joined 18:20 njmurphy left
bartolin_ good evening, #perl6 18:20
leont aristotle: you mean something like always returning a Supply, and middlewares essentially doing .map on those? 18:34
18:39 zakharyas left 18:42 tphilipp joined, tphilipp left
aristotle leont: possibly. not firm enough in Perl 6 to say for sure what to use 18:46
PerlJam But, laziness + composable async primitives for the win in any case. 18:50
18:50 estrabd joined 18:51 Mhoram joined
zostay aritstotle, thx... so can i sum up your thoughts as being, "Why not always return a Promise?" 18:52
aristotle leont: all I can say is the real pain point of PSGI is deferred vs immediate response, whereas streaming body is a side show, and it would be good to remove the inconvenient distinction. I’d question the need for non-/streaming body distinction too, on principle, but I don’t feel strongly about that one 18:53
zostay: yeah
zostay that sounds doable to me 18:54
jnthn A Promise is trivially coercable into a Supply
Which probably helps the streaming thing 18:55
Uh, or uniformly coding against the streaming thing
But letting folks write start { ... } when that's enough
zostay well, the streaming thing is primarily for the content... generally while deferred mostly impacts the delivery of headers 18:56
there are too many adverbs in that last sentence
aristotle and the headers are of much greater interest to other components than the body
b2gills Is there a nicer way of writing `my %h := Hash[Any,Any].new;` or `my %h := :{};` where %h will always be of the same type? 18:57
jnthn my Any %h{Any};
Though the first Any is a tad redundant if you just want the keys to be objects
zostay anyone have a suggestion on what the reference implementation should be called? i have a name, but i don't like it
jnthn Well, psgi is an anagram of pigs, so something about ham or bacon maybe... :P 18:59
b2gills Porki? 19:00
mst for some reason I always end up with the Tool satan song in my head except with baaaacon instead of saaaatan
TimToady what's your kevin psgi number? 19:01
aristotle ouch
zostay lol
TimToady Inversion of Puntrol... 19:02
timotimo mst: UND KEINE SCHINKEN!!
mst timotimo++ 19:04
aristotle zostay: the only thing that comes to mind that’s vaguely in keeping with Rack (Ruby) → Plack (Perl) is maybe Clack (Camelia), which is meh. can’t think of any good way to finnagle “perl 6” or “Six” or something like that into there 19:06
ChoHag Clacks
A la Discworld 19:07
aristotle oh wow I totally missed that
lucasb Clack is the Common Lisp implementation
aristotle lucasb: oh :(
[Coke] Plack6?
19:07 diana_olhovik left, jjido joined
ChoHag Oh now *that's* exciting. 19:07
jnthn Slack? 19:08
(For six...)
zostay that's where i went with it, i want to avoid confusing it with Plack... i ended up with Spackle
RabidGravy +1
TimToady well, how 'bout Plack -> Gingivitis? 19:09
aristotle slack occurred to me but I dunno? tho in a way it’s cool. but it’s going to have a hell of a time competing against the slack.com chat service thing for googlejuice
zostay could name it after the startup script, slackup
TimToady Pl6ck would really confuse the dyslexics
aristotle TimToady: you’re not helping :) 19:10
RabidGravy shouldn't the start up script be call slackoff ;-)
jnthn Heh, "Filling" is what comes after plack...and middleware kinda fills the gap ;P
b2gills I've been trying to come up with a Perl 6 implementation of en.wikipedia.org/wiki/De_Bruijn_sequence that shows off Perl 6's features, but I've got nothing. 19:11
zostay Crack and then the startup script is crackup 19:12
aristotle jnthn: yeah re gingivitis, I was going to say it’s unfortunate enough that “Plack” is the colloquial spelling of (dental) plaque in German
TimToady if you implement with Supply, you could call it Supple 19:13
moritz b2gills: if you post a straight Perl 6 version, I can try to make it more idiomatic
zostay there was an early attempt by someone named Plackdo 19:14
aristotle . o O ( Sixpack? ) 19:15
moritz aristotle: Sixplack?
ChoHag Sixplack
aristotle heh
zostay Cameliack 19:16
[Coke] sixplack++
lizmat this weeks Perl 6 Weekly just got published! p6weekly.wordpress.com/2015/08/03/...ing-along/
19:17 cognominal joined
lizmat publishing a P6W while going at 180 km/hour on the German Autobahn is a special experience 19:17
aristotle how does it feel
b2gills moritz: Part of the problem is I don't really know Python so I'm guessing as it is, I did find one in JS that was easier to figure out en.wikipedia.org/wiki/De_Bruijn_sequence (throw out all code dealing with `memoArr`)
lizmat aristotle: like living in the fast lane :-) 19:18
seems it is my turn to drive
&
19:18 lizmat left
RabidGravy sorry I didn't do any new modules but I started four new ones each more distracting than the last 19:19
ChoHag Put the yaks away. 19:20
RabidGravy but at least I've got a nice pattern for warapping C++ libraries, a lot of ideas for building objects from binary data and some crack induced ideas about ovr-riding enumerations 19:25
moritz today's one-liner: finding all the p6 modules that have a 1; at the end 19:26
19:26 virtualsue joined
ChoHag Wrapping C++ would be nice. People do insist on using it despite it. 19:26
moritz moritz@hack:~/p6/perl6-all-modules$ git ls-files|grep -E 'pm6?$' | perl6 -ne '.say if slurp($_) ~~ /1\;\s*$/'
leont People use C++ because there's not much alternative in its genre. D is largely a cleaned-up C++, but it doesn't have the same momentum or exosystem 19:27
[Coke] leont: that sounds… familiar. 19:28
leont It does
timotimo .o( it wouldn't be terribly bad if what we have was called Frack )
RabidGravy :) yeah but the hippies might start a camp outside worried we're poisoning the water 19:29
ChoHag I killed the debugger. 19:30
Unhandled exception: ctxlexpad needs an MVMContext
RabidGravy but you did'nt shoot the sheriff
ChoHag Nor deputy.
Don't wanna fuck with LEOs these days. 19:31
19:31 yqt joined
ugexe in america it is safer to just shoot the LEO 19:31
19:33 pdcawley left
rjbs on the contrary, SpaceX has been having a lot of trouble shooting things into LEO 19:35
timotimo m) 19:36
what's that other meaning of LEO?
oh
law enforcement officer?
itz low earth orbit
timotimo i was asking for "that other meaning" 19:38
i know about low earth orbit :)
itz ah
RabidGravy leprechaun emerency organization
itz leo sayer?
timotimo blogs.perl.org/users/aaron_baugher/...erl-6.html ← can somebody send a comment this person's way that .pick isn't mutating on the array you're using, and perhaps suggest using .pick(*) to get a shuffled list of the cards to shift or pop from?
dalek ast: 1d56978 | usev6++ | S02-types/nested_arrays.t:
Change test for RT #98954

and remove duplicate ticket number
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...l?id=98954
19:39 [Sno] left
timotimo and calling methods "postfix functions" is ... interesting %) 19:40
19:40 darutoko- left
hoelzro does nqp-j have anything like MoarVM's special_return 19:41
hoelzro was thinking of trying his hand at implementing DESTROY on the JVM
19:45 ChoHag left
[Coke] 1,031 RTs open 19:48
b2gills m: gist.github.com/dwendt/5cc5223d4686d0e33209 19:50
camelia rakudo-moar a38b59: OUTPUT«5===SORRY!5=== Error while compiling /tmp/AyW8s4XgOd␤Preceding context expects a term, but found infix = instead␤at /tmp/AyW8s4XgOd:1␤------> 3memoArr =7⏏5 {};␤»
b2gills damn middle click 19:51
m: gist.github.com/b2gills/18d9496d0634bb83cf78
camelia rakudo-moar a38b59: OUTPUT«0 0 0 1 0 1 1 1␤a a b a c a d b b c b d c c d d␤a a a bb a a ccc a a dddd a bb bb a bb ccc a bb dddd a ccc bb a ccc ccc a ccc dddd a dddd bb a dddd ccc a dddd dddd bb bb bb ccc bb bb dddd bb ccc ccc bb ccc dddd bb dddd ccc bb dddd dddd ccc ccc ccc dddd…»
jjido b2gills: nice! 19:54
b2gills: what does [^$k] do? 19:55
b2gills Array.new( 0 ..^$k )
Array.new( Range.new( 0, $k, :excludes-max )) 19:56
19:56 dha left
jjido b2gills: all right got it 19:57
b2gills: and "!%%" ?
lucasb not divisible :) 19:58
jjido makes sense 19:59
b2gills I really don't like the sub `db` in the middle 20:00
[Coke] RT #121145 is closable with tests if someone wants some karma. :) 20:01
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=121145
20:01 llfourn joined
jjido b2gills: I see what you mean but that's the algorithm. You could pass @a and @sequence as arguments 20:02
skids still finds it strange that with all the crazy operators we never got something for ($a - $a % $b) and the ciel version thereof.
jjido $a -% $b 20:04
[Coke] m: say 20 -% 10
camelia rakudo-moar a38b59: OUTPUT«5===SORRY!5=== Error while compiling /tmp/1uM5pz_IsP␤Two terms in a row␤at /tmp/1uM5pz_IsP:1␤------> 3say 20 -%7⏏5 10␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statem…»
jjido [Coke]: see, it could be added 20:05
[Coke]: btw the answer is 20
20:06 llfourn left 20:07 jjido left
skids "set phasers to stun." "Captain, there appears to be a parrot in my phaser" 20:14
[Coke] didn't we just talk about chaining ~~ ? 20:15
r: say 0 ~~ 0 ~~ 0;
camelia rakudo-{moar,jvm} a38b59: OUTPUT«False␤»
[Coke] masak: if the current state of ~~ and chaining is settled, maybe RT #81336 can be rejected? 20:16
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...l?id=81336
ShimmerFairy [Coke]: it was decided that ~~ is chaining after all (since it was determined that learning that ~~ chains is less disruptive than changing its precedence) 20:19
RabidGravy so I can roll back all the changes that occasioned? 20:20
ShimmerFairy m: say ("foo" ~~ "foo") && ("foo" ge "bar"); say "foo" ~~ "foo" ge "bar"; # these should be the same, so the bug is still valid 20:21
camelia rakudo-moar a38b59: OUTPUT«True␤False␤»
dalek ast: 235a8c9 | usev6++ | S02-types/set (2 files):
Add another two tests for RT #125611
 20:22
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=125611
[Coke] ShimmerFairy: can you clarify on the ticket, then? 20:23
ShimmerFairy [Coke]: well, that's how all chaining ops work, so unless you're referring to something else, I'm not sure what there is to clarify. 20:24
PerlJam ShimmerFairy: I think he means to add that example. 20:25
20:25 brisby joined
PerlJam ShimmerFairy: and if he didn't mean that, I'll ask you to please add your "expanded" chaining example to the ticket. :-) 20:26
20:26 brisby left 20:27 FROGGS left
[Coke] PerlJam++ 20:28
yes, please add that example as something that shoudl work but doesn't. (even better, write a test, and mention in the ticket which file the test went in)
20:29 mr_ron joined
mr_ron m: gist.github.com/ronaldxs/fe83eac02aa690f787d3 20:30
camelia rakudo-moar a38b59: OUTPUT«0␤4␤»
mr_ron Could someone look at the gist - I think it's a bug with Proxy
timotimo huh? 20:31
you mean the 0 in there is problematic?
mr_ron Yes - the zero doesn't make much sense. I don't know that the 4 result makes sense either though ... 20:32
timotimo what did you expect it to print instead, mr_ron?
i think the 0 is absolutely correct 20:33
what else would it be?
mr_ron something like Proxy.new
timotimo oh
well, you'll have to := instead of =
otherwise you'll just fetch the value of that proxy and put it into $p
20:34 rindolf left
timotimo m: my $p := Proxy.new( FETCH => { 1 }, STORE => -> $, $ { say "boing" } ); say $p.perl 20:35
camelia rakudo-moar a38b59: OUTPUT«1␤»
timotimo m: my $p := Proxy.new( FETCH => { 1 }, STORE => -> $, $ { say "boing" } ); $p = 10
camelia rakudo-moar a38b59: OUTPUT«boing␤»
timotimo m: my $p = Proxy.new( FETCH => { 1 }, STORE => -> $, $ { say "boing" } ); $p = 10
camelia ( no output )
timotimo you see? the := vs = makes the big difference
mr_ron I am playing with it now ... thank you very much for your clarification 20:37
ShimmerFairy [Coke]: yeah, I'm thinking of trying to fix it in a couple of minutes, and I'd start by writing tests :) (if I don't start on that soon, I'll just update the ticket) 20:38
timotimo sure, no problem :) 20:40
vendethiel www.reddit.com/r/perl6 <- can whoever has access to this subreddit modding remove the ad? 20:41
ShimmerFairy btw, would it be wise/recommended for me to symlink t/spec to my existing roast repository, so I don't have to deal with going between two copies of the same repo anymore? :)
timotimo i clicked the "remove" button 20:42
is it gone?
seems to be 20:43
mr_ron m: state $p := Proxy.new( FETCH => { 1 }, STORE => -> $, $ { say "boing" } ); $p = 10 20:45
camelia rakudo-moar a38b59: OUTPUT«(signal SEGV)»
timotimo ooooh
20:45 rindolf joined
timotimo yeah, state and := is not compatible 20:45
m: state $test := 1; say $test 20:46
camelia rakudo-moar a38b59: OUTPUT«(signal SEGV)»
timotimo no need for a proxy there, no need for an assignment even
20:46 kaare_ left 20:47 kaare_ joined
mr_ron sorry - I am prematurely optimizing and worrying that every access to a mutator needs to allocate a new Proxy object. 20:48
20:49 brisby joined
timotimo proxies aren't terribly expensive; the more interesting part is probably taking the closures for the fetch and store blocks 20:50
RabidGravy mr_ron, really only need to start worrying about that when you want to do something after the value is set :) 20:51
20:53 llfourn joined 20:54 brisby left
bartolin_ r: my $str; my $i = 0; while $i < 2 { LEAVE { last }; $i++; $str ~= ~$i; }; say $str; $str = ""; $i = 0; while $i < 2 { LEAVE { $str ~= "leaving" }; $i++; $str ~= ~$i; }; say $str 20:54
camelia rakudo-moar a38b59: OUTPUT«1␤1leaving2leaving␤»
..rakudo-jvm a38b59: OUTPUT«1␤␤»
mr_ron m: gist.github.com/ronaldxs/e0a325c9e3bc716cf7f7 20:55
camelia rakudo-moar a38b59: OUTPUT«246␤»
bartolin_ [Coke]: ^^ alas, there still seems to be something wrong on JVM with RT #121145
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=121145
20:55 dha joined
[Coke] bartolin_: hokay. 20:56
jnthn I promised to share whatever progress I'd made on GLR implementation by today. Here it is: gist.github.com/jnthn/aa370f8b32ef98e4e7c9
timotimo oh, GLR interacts with Whatever? 20:57
[Coke] We can at least add a test and make sure we don't regress on moar - don't have to get JVM 100% before christmas.
jnthn timotimo: It interacts with most things :P
bartolin_ [Coke]: I'll add a test for that ... what you said
20:57 kaare_ left
mr_ron the gist is fine. while I have some people looking I was just wondering whether I could get comment on - is the attribute mutator technique a kludge or something likely to work going forward. Technique was displayed yesterday 20:58
20:58 llfourn left
dha Out of curiosity, do we have a command-line option parsing mechanism at this point? I'm not seeing a relevant module (like the Getopt modules in p5) or anything built-in. 20:59
[Coke] dha: sub MAIN
jnthn dha: Look up sub MAIN
timotimo yeah, sub MAIN exists and is pretty great
jnthn mr_ron: There's no plans for Proxy to go away
arnsholt What they said about sub MAIN. Basically, Getopt and friends are obsolete in Perl 6 21:00
timotimo but it has some design decisions in place that you sometimes want to work against
dha Ah, ok. Again, the documentation does not make things easy to find...
21:00 skids left
timotimo pretty often, people get annoyed by having to put positionals in the back and all nameds in front 21:01
jnthn sees we're going from "argh no docs" to "argh I can't find the docs" :)
arnsholt That's an improvement of sorts, I guess. Good point =)
timotimo "too many docs, make them go away!"
mr_ron I didn't mean proxy is going away - the way it is used as an attribute mutator is different from the example in roast which is why I asked - github.com/perl6/roast/blob/master...mutators.t 21:02
dha Which raises a troubling question for me. I know everyone's working hard to get Perl 6 out the door, but... without useful docs, is anyone going to be able to use it?
21:02 Ven joined, Ven left
dha And I want to stress that that question is not intended to start throwing blame around. but it seems like it might be important. 21:02
jnthn dha: Looking at the doc repo, it seems a bunch of people are contributing to the docs too. Heck, I thought you'd done a rather sizable contribution... 21:04
dha I've done a good deal in my own repo. Mostly for changes from p5 to p6 so far. 21:05
jnthn Ah...and planning to migrate that work into doc.perl6.org at some point? 21:06
dha But when I see that things like use and while loops (with the new pointy block stuff) isn't in the main docs, that seems troubling.
jnthn- I would like to, yes. Not sure yet how best to do that. suggestions welcome.
[Coke] dha: yes, it's troubling. Please open tickets on the docs site when you can't find stuff, that'll help. 21:07
github.com/perl6/doc/issues
dha [Coke]- how so? put stuff in Wanted?
ah.
ok.
[Coke] Or in wanted, that's probably fine also.
dha I've added some stuff to wanted. 21:08
21:08 rindolf left
dha I don't want to just turn it into a behemoth that's simply mind-numbing, though. :-) 21:08
jnthn dha: What're you blocking on exactly? Where it belongs, whether it wants a section of it's own becasue it's in many parts, etc? 21:09
dalek ast: 7ce647a | usev6++ | S04-phasers/in-loop.t:
Add tests for RT #121145
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=121145
21:09 TEttinger joined
dha jnthn- The stuff I've done so far might warrant its own section, if only because it's primarily for perl5 people, documenting the changes from p5 to p6. 21:10
[Coke] for sub MAIN, being able to find it via search would be nice - or command line options. I couldn't find it either way. I know there's info in the SYN, though.
jnthn In general, though, please don't suggest that everyone is working on implementation and nobody on docs. It's clear various people are, and I imagine it's a good bit better to motivate them to keep going by acknowledging that. :) 21:11
[Coke] me grabs a copy of doc to see if he can fix the search-didn't-find-anything bug.
dha Although I'm currently also working on a "new stuff in perl 6 that might make you want to use perl 6"
jnthn dha: A "Perl 5 to Perl 6" section sounds sensible enough to me, fwiw. 21:12
dha jnthn- I apologize for making it sound that way. That really wasn't what I was trying to say. I guess I just don't know who's doing docs, and I'm seeing holes in the docs, so I said what I said. In reality, I'm quite sure people are working on the docs, just not in my immediate vision.
Certainly no insult intended. 21:13
[Coke] is seeing lots of Use of Nil in numeric context in block at lib/Panda/Builder.pm:125 21:14
21:15 Ven joined
dha Also, although I don't know how much this holds in general practice, but in my own investigations, I wind up going to the design documents a lot for information, and I don't think (although I could be wrong) that someone trying to learn Perl 6 would go there. Also, true or not, I can't get away from the feeling that the design docs are necessarily what's actually going on in the implementation. 21:15
Again just my, possibly limited, view.
jnthn dha: I was already sure you didn't mean it badly; I've just spent plenty of time listening to "argh why isn't Perl 6 getting done" when I was working hard on the darn thing over the years, and know it can be demotivating. :)
dha And, [Coke] , yes, being able to find MAIN through a search would be helpful. 21:16
jnthn Yes, I agree the design docs aren't where we want people to be looking.
They're certainly in places "plans of where we're heading", whereas docs very much want to be "what's there and how to use it"
dha jnthn- I completely understand that. And if anyone was listening to me and got discouraged, I profusely apologise and open myself to suggestions as to how I can help. 21:17
jnthn- that's how I see it as well.
jnthn Given the amount of work you're doing on 5to6 docs, I'd say you already are helping. Please do record what you find missing, though. Like with bugs in the implementation always going into RT, it's better to have them recorded than forgotten. 21:18
dha I will do my best. 21:19
jnthn dha++
ShimmerFairy Not to mention the synopses are to some extent written with the expectation that you know a thing or two about writing compilers :) (as opposed to simply being a regular Perl 6 user)
timotimo jnthn: liking your gist so far, but i'm not so far into it
dha And if anyone has the time to review what I've written, just to make sure it's accurate, I would be thrilled to put it into the official doc repo.
mst the supplies one? if so, looks *really* good to me
[Coke] mst: new one is the GLR one
mst very much the sort of thing I'd like to see if we could arrange in perl5 if leonerd ever gets certain of his crazy plans to work 21:20
ah
not looked at that one
RabidGravy also doing docs is a strange subset of people in some ways, comfident enough with writing, familiar enough with the subject but not tied up doing stuff to or with the language
[Coke] jnthn++ # Do ALL THE THINGS
21:20 beastd joined
[Coke] wonders if he can make the graphviz stuff go faster when building docs... 21:20
jnthn mst: Glad it looked good to you; I'll be working to get that stuff implemented in the next couple of weeks. :)
timotimo: I think you'll have some fun profiling things. I do have the biases of having written the thing, but the profile output does feel more accessible and easier to work out where the costs are than what's in CORE now... 21:22
21:22 jjido joined, brisby joined
dha And, in case anyone missed the link all the other times I mentioned this... github.com/dha/perl5-to-perl6-docs :-) 21:22
21:22 brisby left
timotimo :3 21:22
dha Also, currently taking suggestions as to features that might make people want to give Perl 6 a try.
tony-o Stage parse : 40.659 21:23
dha: web scale asynchronous solutions
21:24 Ven left
TEttinger the only problem I have with perl 6 is that it's extremely massive and includes a lot of history that I am not familiar with. 21:24
jjido grammars are quite cool.
TEttinger so stuff like, "oh it's just like in perl 5, but X"
doesn't help me at all
b2gills jnthn: Would it make sense for List.elems to ask its iterator(s) about how many elems it thinks it has? For example `(1 xx 100).list.elems` would know that it should return 100 without iterating 21:25
[Coke] if you want the history, then dha's 5-6 is a good place to start. If not, then where we stole the features from probably doesn't matter to you.
TEttinger I don't know perl 5, why would I seek that out?
dha TEttinger - Granted, but I'm guessing a lot of the initial potential audience for p6 are p5 users.
TEttinger I'm guessing a lot of p5 users are now not p5 users 21:26
dha So, yeah, it may not be useful to you, but it may be useful to (many?) others.
tony-o TEttinger has a good point in that isolates just about everyone but p5 users
jnthn b2gills: Do you have a use case?
RabidGravy dha, modern extensible object model, composable concurrency features. consistent design
tony-o who aren't that likely to port existing production code just because
ShimmerFairy dha: I have no problem with docs that help with going from P5 to P6. I think the issue of "don't assume people know Perl 5" is more of a general gripe with P6 documentation (the specs have a bad habit of leaving out "just like P5" stuff in some cases) 21:27
jnthn b2gills: There's going to be a very limited nuber of situations where we could take that shortcut, is all
TEttinger I mean, from what I know about perl 6, which is not much, some of the really good features are that it's extremely good at handling language-like things, with grammars, moar's unicode support, and the best regexes anywhere
21:28 mr_ron left
lucasb b2gills: I had a similar idea about .combinations and .permutations knowing how many elements they have without having to iterate everything :) 21:28
tony-o a grammar tutorial would have been super helpful, froggs and a couple others in here are the only reason i got through the first one i wrote
jjido TEttinger: huh Unicode is a bit of an advanced topic
dha ShimmerFairy- indeed. I am in no way suggesting that the p5->p6 stuff I've done should be the main or, god forbid, only docs. :-)
TEttinger no one else supports 8.0
tony-o a grammar tutorial with actions -
TEttinger java 9 will support 7.0
java 9 comes out next year
jnthn lucasb: OK, now we have an extra use case :) 21:29
flussence grammars are cool, but working with data that isn't perfectly formed utf-8 feels like a massive chore atm
jjido flussence: yeah
ShimmerFairy dha: My only question is how your work will sit alongside the existing 5to6 doc, if your work will go to doc.perl6.org: doc.perl6.org/language/5to6 21:30
leont Has anyone talked to the unicode people about or NFG?
*our
[Coke] ShimmerFairy: I think it will replace it if he can get someone to review it.
ShimmerFairy ah
RabidGravy to be honest I strongly believe that the majority documentation should just treat it as a completely new language with no antecedents
TEttinger RabidGravy: agreed 21:31
tony-o RabidGravy++
b2gills Perhaps someone should go and find things that are in the current 5to6 that aren't in dha++'s version
ShimmerFairy leont: I forget what they called it, but they do have a thing on handling text in more-or-less the same way we do with NFG.
[Coke] RabidGravy: yes, that's the plan.
dha ShimmerFairy - My current off-the-cuff thought is that the Essentials doc I'm currently trying to structure might replace the current 5to6
jnthn Agree, but that doesn't mean we can't have a really give 5 -> 6 guide as well (and heck, a [other language] -> 6 guide)
TEttinger p5->p6 could be in a sidebar even in existing docs, but not assumed knowledge
RabidGravy and I've been doing perl for more than twenty years
[Coke] TEttinger: yes. the official docs will not assume knowledge of p5. the only reason the syn do is because they aren't meant for end users. 21:32
TEttinger ah
dha The other docs I've done have been a bit more exhaustive and probably more useful as reference than tutorial
ShimmerFairy dha: yeah, that makes sense. I was thinking the current one could serve as the introductory doc too, if it's still useful. :)
leont TPF has some sort of membership to Unicode consortium, it may be useful to have one of the p6 core people involved there too (though I suspect the obvious people may be too busy anyway) 21:33
jnthn leont: Not that I'm aware of, but I didn't come up with the idea, just read what Larry'd written in the design docs, asked some questions, read up on Unicode normalization forms, found the most natural algorithmic extension to NFC, and implemented it. :)
b2gills Perhaps there will be other *to6 docs in the future
ShimmerFairy [Coke]: I guess that explains it for the synopses, though someone like me who is both A) interested in the compiler-writing portions of P6, to an extent, and B) doesn't know P5, that doesn't make things better :P
TEttinger 0to6
dha ShimmerFairy - Actually, now that I'm thinking it through, it might be best to leave the current 5to6 sort of as is, but maybe add some stuff to it and make *that* the de factor Essentials doc
jnthn leont: I've got my plate quite full at the moment, as have most of us, but yes, I agree it'd be good if someone from the Perl 6 side was involved there. 21:34
leont: I may be willing, but not for a few months... :)
dha And, yes, the 5to6 stuff would be for a specific audience, and *DEFINITELY* not a replacement for actual p6 docs
ShimmerFairy leont: Well, if I can help out interacting with the Unicode consortium in some way, I did write the initial version of S15, so I guess I might know a thing or two :P 21:35
dha: incidentally, the 5to6 docs help me learn a bit about how Perl 5 does things, which is interesting even if I'll never use it :) 21:36
jnthn ShimmerFairy: I'm curious, did you look over the NFG impl at all? :)
dha ShimmerFairy - An unintended, but useful outcome. :-)
ShimmerFairy jnthn: I don't think I've looked at the code, but I do like what I've tried using so far :) jnthn++ for doing the actual implementation, I'm just the girl who likes to design stuff :P 21:37
ShimmerFairy needs to fix up the rest of the docs to line up with S15 (mainly S05) sometime, now that she's reminded of it. 21:38
timotimo jnthn: if an iterator pushes all it knows how to push, it'll wait for push-at-least to be called again so that it'll return IterationEnd? 21:40
ShimmerFairy dha: > although arrays start with "@", but array I<elements> start with "$", and the like, we've given up.
timotimo push-all returns IterationEnd, so ... ?
jnthn timotimo: It returns IterationEnd also if it knows it ran out
ShimmerFairy dha: I'm not sure I like how defeated that sounds. I think it's for the best we put an end to that :) 21:41
(also, Essentials.pod is missing an =end pod)
jnthn timotimo: It says that in the comment even, no? :)
timotimo ah
jnthn Returns the number of things # pushed, or IterationEnd if it reached the end of the iteration.
timotimo i wasn't 100% sure
fair enough
so when the end is reached, it'll never tell you how much was pushed
jnthn The default impl delegates to push-exactly, which can return IterationEnd
No. Though tbh I don't know of anything that's actually using the number of things pushed yet... 21:42
So we may stop doing that. 21:43
dha ShimmerFairy - Valid points. And I've just tossed some stuff in there kind of randomly so far, so I forgot the =end. Thanks. 21:44
ShimmerFairy dha: in Functions.pod, "A I<lot> of things are objects.": well, _everything_ in Perl 6 is an object, unless the term 'object' means something different in P5. :) 21:46
dha *nod* 21:47
jnthn Time for me to get some rest
timotimo good rest, jnthn
jnthn More GLoRious hacking tomorrow
'night o/
timotimo and thanks for your gist!
dha How about "in perl 6 everything is an object, even though you may not treat them that way in practice" or something along those lines?
Having just taken Damian's "Perl 6: Transparadigm Programming" class, I don't want to give the impression that, even if everything is an object, you have to do things in an object oriented way. :-) 21:48
RabidGravy "may not need to care or need to treat them that way" 21:49
lucasb Differences.pod is so lonely there in the mu repo. Shouldn't it be moved to the doc repo and be merged with 5to6? 21:50
ShimmerFairy dha: yes, I believe S01 or S02 describes it as something along the lines of "everything is an object, but Perl 6 lets you ignore that if you're not interested in using OOP"
smls dha: +1 to expanding the '5to6' doc
if its getting too long, maybe split off separate '5to6-functions' and '5to6-operators' pages 21:51
ShimmerFairy dha: Ah-ha! From S01: "If you want to treat everything as objects in Perl 6, Perl will help you do that. If you don't want to treat everything as objects, Perl will help you with that viewpoint as well."
dha Well put. 21:53
RabidGravy actually a reasonable test of whether that aspiration is met would be to take some complex procedural code and see if it can be ported cleanly 21:54
with the caveat there is a lot of system interaction P6 doesn't do out the box 21:55
21:55 aristotle left
dha Well volunteered, sir! 21:56
ShimmerFairy I think at minimum it means you don't suddenly need to declare the types of all your variables, though if Perl 6 forces you to treat everything as an object in parts, that could potentially be remedied.
(Though I should point out that S01 is just the introductory text to the synopses, more or less, and thus isn't likely to receive updates to line up with modern thinking on P6)
RabidGravy dha, you pick up the samplerate, soundtouch, xbase and soundcloud modules I'm doing and we're good ;-P 21:57
dha Heh. 21:58
ShimmerFairy dha: when it comes to SIGALRM at least, I suggest trying to find an example of intercepting SIGINT (I recall seeing an example of that wrt concurrency a while back). 22:00
RabidGravy being way too distracted the last week
22:02 lucasb left 22:03 dha_ joined 22:04 dha left 22:09 lizmat joined
dha_ ShimmerFairy - Changed the language a bit in Essentials.pod and Functions.pod. thanks for the input. 22:15
ShimmerFairy dha_: I'll be sure to go over the rest of it when I've got time. I hope my odd perspective of "I would actually need a 6to5 document" is helpful :)
dha_ I hope so too. :-) 22:16
22:16 dha_ is now known as dha 22:17 Begi joined, Begi left 22:18 lsm-desktop joined
RabidGravy isn't the common alarm case more like "Promise.in($secs).then({#do something});" seeing as there is no alarm()? 22:26
22:28 skids joined 22:31 smls left, smls joined 22:36 telex left 22:38 telex joined
ShimmerFairy oof, I discovered why smartmatch won't chain, and it's not pretty. Here's another issue that's brought forth by the problem (I'll check to see if it's already rakudobug'd): 22:39
m: class Foo { }; multi sub infix:<~~>(Foo $a, Foo $b) { 42.6 }; say Foo.new ~~ Foo; say infix:<~~>(Foo.new, Foo)
camelia rakudo-moar a38b59: OUTPUT«True␤42.6␤»
ShimmerFairy (basically, the Perl 6 grammar treats ~~ specially, so stuff like this and the chaining bug pop up)
Sgeo Perl gather/take, unlike Python and ES6 and C# etc. etc., take seems to be abstract-over-able, that is, you can hide the take inside a function call. Those other languages let you abstract usage of yield and equivalent, but to use the abstraction you need to syntactically indicate you're doing so, which makes it obvious control is being given up 22:42
You can't abstract over the fact of a yield
I wonder if Perl 6's direction on this is deliberate. I guess personally I prefer explicit. Also you can force a higher-order function to give up control unexpectedly by playing tricks with this 22:43
vendethiel Sgeo: yeah, it's deliberate, it's the same with most other features 22:45
the spec says something along the lines of "if no <keyword> is found in the lexical scope, the dynamic scope will be used"
Sgeo Wait, gather/take mixes lexical and dynamic scope?
vendethiel no. 22:46
Sgeo: quoting the spec "The take function is lexotic if there is a visible outer gather, but falls back to purely dynamic if not. "
It's not mixed, it's one or the other -- can't be both. 22:47
22:47 jjido left
Sgeo So what if I use gather, an inner closure uses take, then I give that closure to something else to run later? 22:47
smls isn't a dedicated gather { } block *more* explicit than resusing a normal function block like in python?
*reusing 22:48
vendethiel m: sub genval(){ take 3; take 4; }; say do given ^3 { genval; }; say ';;;;'; say do given ^2 { genval; }
m: say 1 22:49
camelia rakudo-moar a38b59: OUTPUT«take without gather␤ in sub genval at /tmp/MJoo_RMWSL:1␤ in code at /tmp/MJoo_RMWSL:1␤ in block <unit> at /tmp/MJoo_RMWSL:1␤␤»
rakudo-moar a38b59: OUTPUT«1␤»
RabidGravy gather? not given
vendethiel RabidGravy: whoops; 22:50
Sgeo smls, I'm referring to the sense that you can hide a take behind a function definition so any subroutine might secretly hold a take
vendethiel m: sub genval(){ take 3; take 4; }; say do gather { genval; }; say ';;;;'; say do gather { genval; } # Sgeo
camelia rakudo-moar a38b59: OUTPUT«3 4␤;;;;␤3 4␤»
RabidGravy begins with a g and has an e in it, fair enough
Sgeo vendethiel, now how about if genval was defined in a function that itself had a gather, but was stored somewhere and the later calls called it 22:51
vendethiel m: my $closure; gen gengen { $closure = sub { take 3; take 4; }; }; gengen; say do gather { $closure(); }; say ';;;;'; say do gather { $closure(); } 22:54
camelia rakudo-moar a38b59: OUTPUT«5===SORRY!5=== Error while compiling /tmp/zfuVwer6Vf␤Undeclared routines:␤ gen used at line 1. Did you mean 'get'?␤ gengen used at line 1␤␤»
vendethiel m: my $closure; sub gengen { $closure = sub { take 3; take 4; }; }; gengen; say do gather { $closure(); }; say ';;;;'; say do gather { $closure(); } # >>
TimToady I believe rakudo currently only implements the dynamic form, not the lexotic form
camelia rakudo-moar a38b59: OUTPUT«3 4␤;;;;␤3 4␤»
22:54 llfourn joined, smls left
RabidGravy don't worry I'll be safely retired in fifteen years time ;-) 22:55
Sgeo I think for it to be lexical the literal take would need to be textually inside the literal gather
dalek ast: 18413a7 | ShimmerFairy++ | S03-operators/relational.t:
Add TODO'd tests for RT #81336
 22:56
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...l?id=81336
RabidGravy anyway toodlpip, bed before I turn into a pompkin 22:58
22:59 llfourn left 23:03 RabidGravy left 23:11 virtualsue left 23:16 gfldex left
dha Question: Apparently (according to the docs) "no warnings" doesn't work currently. does "no strict"? 23:17
vendethiel m: no strict; say $a; 23:18
camelia rakudo-moar a38b59: OUTPUT«(Any)␤»
ShimmerFairy m: no strict; $a = 42; say $a; # dha: does this mean 'yes' ? :)
camelia rakudo-moar a38b59: OUTPUT«42␤»
dha I guess so. Debating whether that should be explicitly stated in 5to6, since it does say that "no warnings" doesn't work. 23:19
m: $a = 42; say $a;
camelia rakudo-moar a38b59: OUTPUT«5===SORRY!5=== Error while compiling /tmp/XheqfJL0uN␤Variable '$a' is not declared␤at /tmp/XheqfJL0uN:1␤------> 3<BOL>7⏏5$a = 42; say $a;␤»
dha yep. Looks like that works as I'd expect. 23:20
ShimmerFairy 'no warnings' shouldn't be too hard to implement, I don't think (we got 'no worries' a while back, whose intention is to stop P5 gotcha warnings for those who don't know P5) 23:21
23:22 [Sno] joined
dha I suppose it would be helpful for me to actually learn how things are implemented so I can look and see what is actually going on in the code. But... ow. 23:24
vendethiel wishes #perl6 good night
(you might want to look at "no worries" in the meantime)
(I mean -- it does MORE than just P5 worries right now, IIRC)
dalek kudo-star-daily: b7fa490 | coke++ | log/ (9 files):
today (automated commit)
 23:25
ShimmerFairy vendethiel: yeah, because there isn't a special category for P5 worries or anything. That's why I said "intention", and not anything more concrete :P
vendethiel ShimmerFairy: I actually didn't read your message when I typed mine, and then I was like -- "oops, stale backlog!"
ShimmerFairy and ♞ vendethiel o/
23:26 spider-mario left, vendethiel left
ShimmerFairy dha: it definitely takes a bit of time to get used to reading the Perl 6 grammar as opposed to grammars written in Perl 6 :) 23:26
dha I would imagine. What do I need to be able to read to make sense of the rakudo source? C? nqp? Farsi? 23:27
ShimmerFairy dha: The underlying stuff is written for NQP, which is a subset of Perl 6 specifically geared towards compiler writing. It's only for the vm-specific stuff that you need to know how to read C or Java or what-have-you 23:28
dha Ok. Guess I'll try to find time to learn nqp, then. 23:29
ShimmerFairy dha: There was a two-day course a couple years back that explained how to write a compiler in nqp (talking about rakudo, with exercises to implement your own mini-compiler). I can link you to it if you like.
dha That would be great. Thanks. 23:30
ShimmerFairy dha: here you go :) github.com/edumentab/rakudo-and-nq...als-course
dha Oops. I think I found a mistake in the 5to6 doc. It says that -M and -m have been replaced by --use, but perl6 -h says that -M is how you load a module on the command line. 23:31
ShimmerFairy dha: yeah, I recall being confused a bit by that. (--use might work too, but it certainly hasn't replaced anything) 23:32
flussence `perl6 --use` agrees with `perl6 -h` there (illegal option)
dha Ah. So that needs to be changed. Will look into that. 23:33
ShimmerFairy I was just about to say, turns out --use does nothing in the first place :)
dha And, I'm guessing that the ability to "no" a module on the command line via "perl -M -Perl::Module" is gone? 23:37
23:37 lucasb joined
dha (it is, to my surprise, in Perl 5) 23:37
ShimmerFairy perhaps, I've never heard of that before.
Note that -e turns off strict, if that's what you're going after :) 23:38
dha Oh? That's interesting.
ShimmerFairy I personally don't see why it should, but I chalked it up to another one of those P5 things I'll never understand :P 23:39
dha Huh. When I try "perl6 -M -Linenoise" it starts up the REPL, but when I *exit* it spits at me that "-Linenoise" can't be found. That seems... late.
ShimmerFairy dha: or if you try typing a statement 23:40
dha Also, it doesn't let me actually exit.
flussence would be nice to have a perl6 -E which means "keep strict on", that'd be sorta-consistent with what perl -E does 23:41
dha In any case, I'm going to assume that behavior has changed from p5.
23:41 Peter_R left
lucasb m: say 'abc'.index('a') 23:43
camelia rakudo-moar a38b59: OUTPUT«0␤»
lucasb m: say (<a b c>).combinations('abc'.index('a')).perl
camelia rakudo-moar a38b59: OUTPUT«()␤»
lucasb m: say (<a b c>).combinations(0).perl 23:44
camelia rakudo-moar a38b59: OUTPUT«((),)␤»
lucasb Maybe someplaces will need to change 'if/unless $n' to explicit test for equal/not-equal zero to accomodate for the existence of the new Index type.
dha flussence - -E doesn't turn strict on on Perl 5. 23:45
flussence I know, that's not what I mean though. It turns on "new stuff after 5.8", and here strict-by-default would also be new stuff 23:47
dha Hm... I don't know. In the context of Perl 6, *everything* is "new stuff". :-) 23:49
Zoffix But it's not. Warnings by default, it seems. 23:50
Sgeo Is gather/take magic, or does it use control exceptions or something? 23:54
TimToady yes, control exceptions 23:55
at least in the current implementation
this may change after the GLR, at least in cases where we know it's in an eager context 23:56
Zoffix GLR? 23:57
TimToady Great List Refactor 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!