»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
dataangel p6: class zz { has @.yy is readonly; }; my $xx = zz.new(yy => [1, 2]); say $xx.yy; $xx.yy = [3, 4]; say $xx.yy; # why am I able to reassign it if it is read only? 00:03
camelia rakudo-moar 6cd6ef: OUTPUT«[1 2]␤[3 4]␤»
gfldex should .f on a dangling symlink throw an exception? 00:04
gfldex the consequence is that any program that is working with files has to special case dangling symlinks 00:05
dataangel Why bother having classes and roles instead of just roles? They seem strictly better, you get errors instead of confusing behavior 00:11
jdv79 roles can't do a bunch of things. they are fundamentally different. 00:15
timotimo is there some context to that role vs class thing you just said? 00:18
we might want to emit errors or warnings for "is readonly" on @ or % sigiled attributes
dataangel timotimo: roles give you errors when you have conflicting method names, and apparently classes don't 00:19
timotimo readonly will only cause a getter (not a getter-setter) to be generated
but that's only about changing the value of "outer" scalar containers
you can only have conflicting method names when you do multiple inheritance
otherwise inheritance has the explicit feature to replace a method with the same name by another
though we may want at some point to have something like java's @Override that forces you to say which class you override a method from 00:20
dataangel timotimo: oh you can't do virtual's with roles
timotimo on the other hand, you can still always $theThing.Class::method(...)
you totally can
but a role that gets "does"ed into a class acts just like if you had just written the contents of the role { ... } into the class {...} itself
dataangel timotimo: what is the purpose of only generating a getter if you can reassign the array anyway?
timotimo it's not really re-assigning 00:21
it's "storing into"
we don't have "deep read-only-ness" or "deep immutability" anyway, because that's *really* hard 00:23
pi____ m: class Account-Holder { has $!name; my class Account { has $!account-number; } } 00:35
camelia ( no output )
pi____ how do i access the account number in the above code? 00:36
timotimo there are multiple ways. one of them is to make it "our class Account" and go Account-Holder::Account
another is to build a little accessor that does nothing but return Account
pi____ would that accessor be a method inside the Account class? 00:37
timotimo i'd imagine so 00:38
methods don't have to be called on an instance, if they don't access attributes of self
pi____ hmm, could i bother you for an example?
timotimo m: class Account-Holder { has $!name; my class Account { has $!account-number; }; method get-acc-class { Account } }; my $foo = Account-Holder.get-acc-class.new(:account-number(9999999)); say $foo 00:39
camelia rakudo-moar 6cd6ef: OUTPUT«Account-Holder::Account.new␤»
timotimo ah, private attributes don't get set by the default constructor, nor are they output by the default .perl
i'm about to go home so i can sleep :) 00:41
pi____ all right haha thanks
timotimo have fun with perl6 :) 00:41
pi____ you bet!
lambd0x m: my @a = <1 2 3 4>; sub no ([$pivot, $pivot2, *@a]) { say $pivot ~ " " ~ $pivot2;} ; say no(@a); 00:54
camelia rakudo-moar 6cd6ef: OUTPUT«1 2␤True␤»
lambd0x that's so nice 00:55
Guys does the * here turns @a into a pointer from which I take the first, second... element for the sub? 00:56
gfldex it's the rest 01:03
grondilu is reading perl6.party/post/The-Awesome-Errors-of-Perl-6 01:04
m: say "foo" . "bar";
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at <tmp>:1␤------> 3say "foo" .7⏏5 "bar";␤»
grondilu ^seems a bit excessive to assume that strongly that a concatenation was intended. 01:05
timotimo m: sub infix:<.>($a, $b) { $a ~ $b }; say "hello" . "world"
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at <tmp>:1␤------> 3x:<.>($a, $b) { $a ~ $b }; say "hello" .7⏏5 "world"␤»
timotimo m: sub infix:<.>($a, $b) { $a ~ $b }; say "hello"."world"
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Quoted method name requires parenthesized arguments. If you meant to concatenate two strings, use '~'.␤at <tmp>:1␤------> 3$a, $b) { $a ~ $b }; say "hello"."world"7⏏5<EOL>␤»
timotimo hm. i guess. 01:06
grondilu I would have said something like "undefined use of '.' as a binop between string. In Perl 6, string concatenation is done with ~."
timotimo i think it should be possible to override .
grondilu or seomthing like that.
timotimo "between strings" isn't all, though 01:07
as you can use string concat to stringify, too
grondilu yeah.
grondilu maybe just add a "presumably" adverb. 01:09
grondilu like: "Unsupported use of ., presumably to concatenate strings." 01:10
I would agree that it's no big deal at all though. The current message is not too bad. Just bugged me slightly. 01:11
SmokeMachine____ how can I add elements on a SetHash inside of a hash? 01:28
m: my SetHash %a; %a<new_sethash><str_element_of_sethash> = True; %a.perl.say 01:29
camelia rakudo-moar 6cd6ef: OUTPUT«Type check failed in assignment to %a; expected SetHash but got Hash (${})␤ in block <unit> at <tmp> line 1␤␤»
timotimo it can't automatically create the SetHash for you, i'm afraid
maybe that's buggable
m: my SetHash %a; %a<new_sethash> .= new; say %a.perl 01:30
camelia rakudo-moar 6cd6ef: OUTPUT«(my SetHash % = :new_sethash(SetHash.new()))␤»
timotimo m: my SetHash %a; %a<new_sethash><a> = 1; say %a.perl
camelia rakudo-moar 6cd6ef: OUTPUT«Type check failed in assignment to %a; expected SetHash but got Hash (${})␤ in block <unit> at <tmp> line 1␤␤»
timotimo m: my SetHash %a is default(SetHash); %a<new_sethash><a> = 1; say %a.perl
camelia rakudo-moar 6cd6ef: OUTPUT«Type check failed in assignment to %a; expected SetHash but got Hash (${})␤ in block <unit> at <tmp> line 1␤␤»
timotimo so autovivification apparently doesn't work with typed hashes
timotimo so you'll have to //= .new before accessing a sethash-inside-the-hash to make sure it exists 01:31
well, //= bleh.new
SmokeMachine____ m: my Hash of SetHash $a .= new; $a<bla><ble>++; dd $a
camelia rakudo-moar 6cd6ef: OUTPUT«Hash[SetHash] $a = (my SetHash % = :bla(${:ble(1)}))␤»
SmokeMachine____ but as a scalar it works... why? 01:32
m: my Hash of SetHash $a .= new; $a<bla><ble> = "not a possible set value"; dd $a
camelia rakudo-moar 6cd6ef: OUTPUT«Hash[SetHash] $a = (my SetHash % = :bla(${:ble("not a possible set value")}))␤»
SmokeMachine____ now that doesn't make sense... 01:32
timotimo that's pretty weird 01:33
looks to my tired eyes like a rakudobug
SmokeMachine____ timotimo: the scalar and the hash versions should do the same, right? 01:35
timotimo not sure 01:37
you can see if the generated code is the same with --target=ast :) 01:38
i can see if my pillow is still on my bed with --target=sleep
o/
SmokeMachine____ :) have a good night! 01:39
timotimo have a good * :) 01:40
gfldex m: my (Bool :$foo, Bool :$bar) = True; dd ?$foo, ?$bar 01:49
camelia rakudo-moar 6cd6ef: OUTPUT«Bool::True␤Bool::False␤»
sammers ola perl6 01:53
sammers are there any examples of capturing keystrokes as a Supply? 01:58
gfldex sammers: you need to set the terminal to raw mode what may clash with concurrency 02:00
see github.com/krunen/term-termios
sammers gfldex, thanks I will take a look. I have been messing around with this example, trizen.gitbooks.io/perl6-rosettaco...acros.html 02:02
AlexDaniel .seen raiph 02:15
yoleaux I saw raiph 13 Aug 2016 18:25Z in #perl6: <raiph> dataangel: Most Perl 6 string and numeric types implicitly coerce between string and numeric interpretations by default; you have to be explicit if you don't want that
AlexDaniel .tell raiph Using readable names for placeholder values (like $^key and $^value) is a footgun, please consider not recommending it publicly 02:17
yoleaux AlexDaniel: I'll pass your message to raiph.
SmokeMachine____ m: Str(Numeric) ~~ Int 02:20
camelia ( no output )
SmokeMachine____ m: say Str(Numeric) ~~ Int 02:21
camelia rakudo-moar 6cd6ef: OUTPUT«False␤»
SmokeMachine____ Shouldn't be true? Why?
m: say Int ~~ Str(Numeric) 02:22
camelia rakudo-moar 6cd6ef: OUTPUT«False␤»
SmokeMachine____ m: say Str(Int) ~~ Int 02:23
camelia rakudo-moar 6cd6ef: OUTPUT«False␤»
SmokeMachine____ m: say Str(Int) ~~ Str 02:24
camelia rakudo-moar 6cd6ef: OUTPUT«False␤»
gfldex sammers did you manage to get the terminal into raw mode? Seams not to work for me. 02:27
sammers: got it :) 02:28
gfldex sammers: that's how far I got gist.github.com/60ec413cc652bbbc6e...b413338237 02:51
timotimo does start { ... }.Supply really work like that? i don't think so. 02:52
timotimo i think you're just turning the promise into a supply that'll emit exactly once when the promise is kept 02:53
i.e. when you've entered 10 keystrokes, you'll get a result from the started block
i'd expect the emit in that case to fall through to the outermost react { ... } that your stuff is in
and thus be completely ineffective for the whenever loop you've meant to use it for
but i think it should error out instead 02:54
it doesn't have a supplier in its dynamic environment, i don't think
back to trying to get to sleep ...
BenGoldberg m: say Str(Int).WHAT; 03:05
camelia rakudo-moar 6cd6ef: OUTPUT«(Str(Int))␤»
gfldex sammers: this works: gist.github.com/60ec413cc652bbbc6e...b413338237 03:25
Lamjarred Gabor Szabo? 03:54
hello all
dalek pan style="color: #395be5">perl6-examples: 22ed552 | (David Warring)++ | t/categories/parser (2 files):
rename t/categories/parser-SimpleString.t -> t/categories/parsers/SimpleStrings.t
 04:18
pan style="color: #395be5">perl6-examples: 2300b77 | (David Warring)++ | / (2 files):
add some basic CSSGrammar tests. Fix `@page` rule
melezhik Hi all! 09:45
DrForr Howdy!
melezhik A hopefully simple question on perl6 subroutines
given a perl6 module Foo::Bar
melezhik and I want to call a subroutine foo which is defined at this module 09:46
by "full" name
melezhik like in perl5 I will have something like Foo::Bar::foo() 09:46
how can I do this in perl6?
also, my module gets used in runtime 09:47
DrForr m: class Foo::Bar { sub bar { say "hi" } } Foo::Bar::bar();
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Strange text after block (missing semicolon or comma?)␤at <tmp>:1␤------> 3class Foo::Bar { sub bar { say "hi" } }7⏏5 Foo::Bar::bar();␤ expecting any of:␤ infix␤ infix stoppe…»
DrForr m: class Foo::Bar { sub bar { say "hi" } }; Foo::Bar::bar();
camelia rakudo-moar 6cd6ef: OUTPUT«Could not find symbol '&bar'␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
melezhik sub module_run($name, %args = %()) is export { my $mod_name = 'Sparrowdo::' ~ $name; require ::($mod_name); }
so I can't hardcode a name of module as it resolved in runtime 09:48
so having "require ::($mod_name); " i want to call some $mod_name's subroutines 09:49
and last thing it should be module, not class
gfldex melezhik: if you load a module at runtime you have to do a runtime lookup for the symbol
melezhik yes, sure 09:50
I know how to do this in perl5
I just need a hint on perl6 syntax
gfldex ::("YourClassHere") 09:50
tadzik require 'Foo'; ::('Foo').symbol
melezhik tadzik. I can't hardcode a Foo; as Foo is in $mod_name 09:51
gfldex ::("$YourClassHere") 09:52
TimToady DrForr: you need an 'our' for that to work
yoleaux 15 Aug 2016 01:21Z <gfldex> TimToady: you may want to spend some attention on m: sub f { 42 }; my ($i where * == 2) = f; (backlog: irclog.perlgeek.de/perl6/2016-08-15#i_13023085)
DrForr 'our sub... ' yeah, forgot. As usual, doing more things in the background.
DrForr Also, (Terry Riley)++ 09:53
tadzik melezhik: yes, that's why I used a string in there in both cases :)
so in your case, require $mod_name; ::($mod_name).blabla
melezhik tadzik : can I have this? - :($mod_name).$funcname ? 09:54
not sure ... really
tadzik in that case you may need ::("$funcname::$mod_name") .....I think 09:55
melezhik ahhh . funcname is defined and could be a constant ... so you example will be fine
tadzik I always used that for methods
melezhik tadzik. will try and let you know, thanks
melezhik tadzik, looks like your way relates to class, not module; my case: use v6; unit module Sparrowdo::Nginx; use Sparrowdo; sub tasks { .... }; 10:10
I have an error - Method 'tasks' not found for invocant of class 'Sparrowdo::Nginx'
when try this code - $name = 'Nginx'; my $mod_name = 'Sparrowdo::' ~ $name; require ::($mod_name); ::($mod_name).tasks 10:11
tadzik right, how about the one I suggested later?
melezhik what exactly? 10:11
sorry, if missed something ...
arnsholt The $thing.method is purely for invoking methods 10:13
What tadzik is referring to is `::$("$mod_name::$func_name")` I think 10:14
Which probably also requires you to make it `our sub tasks { ... }`
If you want to make the methodcall syntax work, you could make it `unit class` and `method tasks` 10:15
melezhik hm . I don't want a methods, I want plain subroutines
what about this?
gregf_ m: class Foo::Bar { method bar() { say "Bar!" } }; say Foo::Bar.bar # should this work melezhik ?
camelia rakudo-moar 6cd6ef: OUTPUT«Bar!␤True␤»
melezhik I don't want call this as class methods, while it's doable, but I need I simple subroutines
it probably will work, but what about calling some module's function, using Foo::Bar::name style , like I would have in perl5? 10:16
nine melezhik: just do it? 10:17
melezhik: it's exactly the same syntax in Perl 6
melezhik just do what?
ahh, ok
gregf_ well, it doesnt have to be just like in Perl5 *literally* - *smiles*... Perl6 is a new language ;)
nine m: module Foo::Bar { our sub foo { say "hello"; } }; Foo::Bar::foo; 10:18
camelia rakudo-moar 6cd6ef: OUTPUT«hello␤»
melezhik yeah, this is why I told style, not way ... )))
gregf_ melezhik: Foo::Bar->bar() # is how youll do this in moose as well,
nine has the solution! next question :) 10:19
melezhik ahhh, guys, again I don't want to call it like class method .... ok, I will try a Foo::Bar::func ... and will let you know ... 10:19
nine melezhik: also why exactly is the module loaded at runtime?
melezhik ok, I see nine's example will try it too ... 10:19
tadzik yep, I meant what arnsholt++ said :) 10:21
melezhik nine, it's about upper API style, I don't want users will "use Some::Module" and just have some other syntax, but probably I will need to go a "standard" way, getting use modules explicitely ...
nine melezhik: Sorry, I don't think I understood what you're trying to say? 10:22
melezhik nine I mean I use a runtime module loading for the sake of simplicity of upper code relying in the modules, but nevermind now ))) 10:24
arnsholt I suspect using classes will give you a cleaner inheritance and composition story, even if the classes aren't instantiable/don't have any instance methods 10:24
sammers hi perl6 10:25
DrForr Afternoon, sammers
arnsholt Since it lets you provide default implementations for trivially implemented methods and the like
DrForr (at least locally)
arnsholt G'noon =)
sammers afternoon, yeah, here it is ~19:30
so, I am trying to make a single react / whenever respond to multiple supplies, is this possible? 10:26
jnthn whenever Supply.merge($s1, $s2, ...) { }
sammers ah, great 10:27
melezhik arnsholt I got your point, the code is pretty raw , I am at the begining of the desing, will take your words into account ...
arnsholt jnthn++ to the rescue!
TheLemonMan hmm, when is it advised to use sigilless parameters for functions? from what I glean from the documentation it is another "lighter" way to pass readonly content
timotimo sigilless is equivalent to "is raw" 10:29
jnthn When you don't want to enforce any context on the argument, generally 10:30
TheLemonMan oh, I didn't get you'd completely lose the context. Thanks for the heads-up 10:31
timotimo you don't lose it, yo ujust don't force any context on the argument 10:32
jnthn m: sub foo($a) { for $a { .say } }; foo([1,2,3])
camelia rakudo-moar 6cd6ef: OUTPUT«[1 2 3]␤»
jnthn m: sub foo(\a) { for \a { .say } }; foo([1,2,3])
camelia rakudo-moar 6cd6ef: OUTPUT«[1 2 3]␤»
jnthn Um...well that didn't do what I expected :P 10:33
oh, duh
m: sub foo(\a) { for a { .say } }; foo([1,2,3])
camelia rakudo-moar 6cd6ef: OUTPUT«1␤2␤3␤»
jnthn It'll only do what I expect if I type it right :P 10:34
Anyway, the $a means "this is an item", so when you iterate you're just going to get a single iteration
Whereas \a is non-commital. If the original thing would iterate as many items then that'll happen.
TheLemonMan hmm, I'm slightly less confused now heh 10:38
melezhik finally I decided to go a class way and it works. thansk to all! 10:41
tadzik so the ::("$foo::$bar") syntax didn't?
melezhik this is what I get when try to this in none class way: 10:48
::('Sparrowdo::Nginx::tasks')(%args)
unit module Sparrowdo::Nginx; sub tasks () {...} 10:49
and error is:
No such symbol 'Sparrowdo::Nginx::tasks'
timotimo i'd think you'd have to manually split on :: and use the :: operator each time?
timotimo ::('Sparrowdo')::('Nginx')::tasks 10:49
melezhik well, let me try
melezhik the same error - No such symbol 'Sparrowdo::Nginx::tasks' 10:51
when try to invoke via - ::('Sparrowdo')::('Nginx')::tasks(%args);
timotimo OK 10:52
melezhik head lib/Sparrowdo/Nginx.pm6 - use v6; unit module Sparrowdo::Nginx; use Sparrowdo; sub tasks (%args) {
timotimo well, for one, you could also type that Sparrowdo::Nginx::tasks :)
and only make the parts that are dynamic use the ( ) 10:53
arnsholt I experimented with camelia and couldn't get it to work
timotimo OK
arnsholt m: module Foo { our sub foo { say "foo" } }; Foo::foo() # works
camelia rakudo-moar 6cd6ef: OUTPUT«foo␤»
timotimo i don't know much about this part of things
arnsholt m: module Foo { our sub foo { say "foo" } }; ::("Foo")::foo() # nope? 10:54
camelia rakudo-moar 6cd6ef: OUTPUT«No such symbol 'Foo::foo'␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
arnsholt Rakudobug perhaps?
melezhik arnsholt: let me try ...
timotimo m: our module Foo { our sub foo { say "foo" } }; ::("Foo")::foo() 10:55
camelia rakudo-moar 6cd6ef: OUTPUT«No such symbol 'Foo::foo'␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
timotimo m: our module Foo { our sub foo { say "foo" } }; ::("Foo")
camelia ( no output )
timotimo m: our module Foo { our sub foo { say "foo" } }; ::("Foo")::.keys.say
camelia rakudo-moar 6cd6ef: OUTPUT«()␤»
timotimo m: our module Foo { our sub foo { say "foo" } }; ::("Foo")::('&foo')
camelia ( no output )
timotimo that's the magic ingredient
you need the sigil in there, as it's part of the name
not sure why it doesn't show up in the .keys, though
melezhik yeah, arnsholt: you had it right, I need an our sub ... 10:55
"our" is vital part ... 10:56
timotimo without "our" it won't be accessible via the package at all
for subs, the default is "my", which means "only available in the lexical scope"
melezhik yeah, now I see ...
CIAvash Am I the only one who thinks our errors don't look good when you compare it to Rust and Elm's? 10:57
timotimo isn't elm one of the haskell-to-javascript languages?
CIAvash yes 10:57
well, not exactly haskell, it has similarities 10:59
ufobat ola p6 :-) 11:01
melezhik unfortunately this code still does not work: my $func_name = 'Sparrowdo::' ~ $name ~ '::tasks'; ::($func_name)(%args);
No such symbol 'Sparrowdo::Nginx::tasks' 11:02
not sure what I had wrong ...
my module code is - use v6; unit module Sparrowdo::Nginx; use Sparrowdo; our sub tasks (%args) { ... }
timotimo you forgot to put the & before tasks
bjz timotimo: it's like haskell, but with all the astronautical type system stuff ripped out and a simpler syntax (ie. less syntactic sugar) 11:03
melezhik timotimo - works! thanks a lot! 11:03
)))
timotimo yay
melezhik so I did a subroutine/modules way too! 11:04
at least I have to ways by hand now )))
CIAvash look at this comment for example, "The second error needs multiple paragraphs to explain. Why isn't it instead something like:" www.reddit.com/r/programming/comme..._6/d6j42wh 11:05
melezhik arnsholt: - thank you too! 11:06
timotimo CIAvash: yeah, we should really put the eject sign at the beginning of the block, not at the end 11:07
CIAvash timotimo: it's not just the eject sign. the amount of code to show in the error, the explanations and where they appear 11:08
melezhik a final working version is - sub module_run($name, %args = %()) is export { require ::('Sparrowdo::' ~ $name); ::('Sparrowdo::' ~ $name ~ '::&tasks')(%args); }
TheLemonMan I don't quite like the eject sign tbh 11:09
probably the fact that I never configured urxvt font fallbacks correctly concurs to that 11:10
CIAvash I also like the way Elm and Rust show the line numbers 11:11
JIC someone has not seen it yet: blog.rust-lang.org/2016/08/10/Shap...-come.html elm-lang.org/blog/compiler-errors-for-humans 11:15
also, is there a reason run-time errors don't show the code in the error? 11:18
timotimo you have to set an env var if you want that 11:19
RAKUDO_VERBOSE_STACKTRACE or something along those lines
CIAvash oh
lizmat clickbaits p6weekly.wordpress.com/2016/08/15/...-go-wrong/ 11:22
CIAvash: the value of RAKUDO_VERBOSE_STACKFRAME is the number of lines before/after you want to see 11:25
CIAvash thanks 11:28
DrForr lizmat: Feeling better?
lizmat DrForr: yeah, it's mostly nagging bruises now 11:29
lizmat I never thought I would have green fingers, but now I have 11:29
DrForr Good to hear. (see, actually)
That sounds serious. Ow. 11:30
lizmat oddly enough, no trouble with typing...
DrForr I was wondering about that...
lizmat but picking up heavier things with my left hand is still too painful
AlexDaniel lizmat: sounds great! According to my linear predictions, csv tests should be under one second now… so please catch up! 11:31
on a slightly more serious note, I'm glad that you feel better :)
lizmat AlexDaniel DrForr thank you for your kind words :-) 11:32
timotimo lizmat: i'm hoping for speedy, complication-free, and effortless recovery :)
AlexDaniel
.oO( compilation-free? )
 11:33
lizmat timotimo: thank you
timotimo hah
unmatched} CIAvash: TBH, I feel like people are putting these examples on a pedestal and showing them off without proper context. The problem with all those "^^^^ this bit blah blah" IMO is the errors are always different. I have to scan for the "^^^" bits amid my code. Whereas if the error always looks the same, I can just look at the spot where the line number is or where the short code sample is. If I see ".^name" 11:42
(the carret makes it easy to spot), I already know it's the unitialized Str warning and I can just skim to the end of the line to see the line number. It wouldn't be so easy with giant code samples in the way. Also, I wonder how all of that ASCII-art-in-code style looks with real-world code rather than crafted code samples. Some of the lines in Rakudo source are over 80 chars, much over. If I have an error at
the end of such a line and the error displayer shoves a long error at the end of it, we're looking at a line that won't even fit into my terminal.
El_Che I am trying to add perl6 support to vim syntastic (a linter) and a must say the free form of errors has been trying for my limited vimscript skills 11:44
unmatched} regrets writing that article now 11:45
jnthn El_Che: But most errors are structured objects
unmatched} We have bigger problems to focus on right now than what trolls on Reddit think about our error messages.
jnthn El_Che: Which you could dump out in some easy-to-handle form (JSON or so)
arnsholt jnthn: Which are then passed as unstructured text into the next stage of the pipeline >.<
jnthn Wat? 11:46
Which pipeline? :)
timotimo the user's terminal
in this case, at least
arnsholt Like in El_Che's vimscript example, the raw string of the error message is passed as-is into the linting stuff
Which is, as you rightly point out, not a very good idea 11:47
I guess making Rakudo capable of dumping exceptions as JSON shouldn't be too onerous?
El_Che oh, maybe I hijacked the wrong conversation
El_Che let me explain 11:48
arnsholt An env-var/command-line parameter to toggle it, and some code in the top-level exception handler? 11:48
jnthn Yes, my point was that you would probably be better off writing a small wrapper script that compiles the code to lint, CATCHes the compilation exception, and dumps it in some standard form :)
El_Che I loved Perl6 errors. They are clear for the user
jnthn Then you can deal in type names and pull out structured info 11:48
andreoss can i forbid .new to be called on class?
El_Che (terrible wifi here, let me read the backlog)
jnthn method new(|) { die "Can't new this" } 11:49
timotimo can't newt his what?
ilmari is it really a class if you can't .new it?
arnsholt jnthn: Yeah, exactly. My background in NLP just makes me particularly allergic to the traditional Unix way of flinging unstructured text around and expecting robust software to somehow magically occur =)
El_Che I kind of got most of the syntastic perl6 support working, but it's fragile as it indeed parses the "perl6 -c" output 11:50
arnsholt andreoss: `method new { die "Nope!" }` should do the trick, shouldn't it?
El_Che and some errors have the same pieces of information in an other place
andreoss jnthn: `subtype Null is Foo; # singleton `
arnsholt Alternatively, is repr('Uninstantiable')
andreoss this would not be correct?
El_Che so a json error switch would be a wow
CIAvash unmatched}: I just referred to one part of the reddit comment. Blog posts and discussions sometimes can lead to good things, don't regret it :) 11:51
jnthn I don't think you can .new on a subset type either, so that's another way
andreoss m: class Nil3 is repr('Uninstantiable') { }; Nil3.new.say
camelia rakudo-moar 6cd6ef: OUTPUT«You cannot create an instance of this type (Nil3)␤ in block <unit> at <tmp> line 1␤␤»
andreoss *subset 11:51
El_Che also similar on the perl5 linter (perl -c), the perl 6 one will be disabled out of the box because of security concerns regarding -c 11:53
nine ilmari: it can still have other constructors beside .new. Also it could be a static class. 11:54
gfldex unmatched}: if you would have written the same blog post about rust, you would have seen the same amount of hate. There are always folk who feel threaten by change but wont express the reason for their irrational behaviour. 11:55
gfldex unmatched}: assholes have a function in human society, they teach use humility. 11:57
andreoss m: class Foo { has $.a } ; subset Bar is Foo where *.a ~~ Int;
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Method 'add_parent' not found for invocant of class 'Perl6::Metamodel::SubsetHOW'␤at <tmp>:1␤»
andreoss is this supposed to work?
timotimo no 11:58
it has to be "subset Bar of Foo"
gfldex LTA tho
timotimo if you just write "subset Bar", you get a subset of Any. then you say "it's derived from Foo"
yeah, the error could be made clearer by a bit of special-case knowledge
thing is, class derivement is just one case of "is" 11:59
and it stands to reason you want to add some traits to your subsets
if add_parent doesn't exist, the "is" trait for derivation could say "you cannot derive a SubsetHOW from a class" 12:00
thought that sounds like "you can't make subsets out of types" which is kind of bogus
timotimo and we'd have to run to the end of the statement to figure out if there's an "of Bar" that comes after the "is Bar" 12:00
to figure out if the user left out the "of" part by accidentaly writing "is" instead 12:01
*shrug*
i'm sure someone will find a cleverer way than what i'm thinking of :)
andreoss is it not possible to write somehthing like `subset Foo(Int $a, $b) of Bar;`? 12:03
gfldex unmatched}: thinking about it, you did make one mistake. You forced the opinion that those errors are awesome on the reader - in the very first line. You may want to leave it to them to make that judgement.
andreoss: what would you want the signature do in that case? 12:04
andreoss gfldex: attribute list 12:07
lizmat andreoss: we even have an exception for it: X::Cannot::New 12:08
andreoss i thought it would be posible to use subsets as ADTs. 12:10
`role List {}; subset Nil of List; subset Cons($a, List $b) of List;`
m: subset Foo($a); 12:11
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Variable '$a' is not declared␤at <tmp>:1␤------> 3subset Foo(7⏏5$a);␤»
gfldex Sir, please lay down that Nil! 12:12
andreoss so it's legal to have singature in subset definition? what can i put there?
gfldex m: my $a; subset Foo($a); 12:13
camelia rakudo-moar 6cd6ef: OUTPUT«Cannot invoke this object (REPR: Uninstantiable; Foo)␤ in block <unit> at <tmp> line 1␤␤»
jnthn andreoss: No, subset types are refinement types 12:14
They take an existing type and add a condition (the "where")
gfldex being not-illegal doesn't mean it works.
jnthn All that's doing is invoking the subset 12:15
Which doesn't do anything useful :)
gfldex m: subset Foo of Str; sub f(Foo){}; &f.signature.params[0].say 12:16
camelia rakudo-moar 6cd6ef: OUTPUT«Str $ where { ... }␤»
gfldex there is no introspection for subsets directly, but you can get there via a signature
unmatched} gfldex: nah, the title was deliberately provocative :) My regret is not the reddit comments themselves but I have two Perl 6 channels open and both of them are discussing Perl 6's errors based on commentary of those reddit trolls, and some people even proposing we start opening RT tickets for a whole category of error "LTA-ness". 12:17
I feel that's a distraction.
jnthn gfldex: You can introspect the refinement and the refinee using the MOP 12:18
m: subst Foo of Str where *.chars > 10; say Foo.^refinee
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Two terms in a row␤at <tmp>:1␤------> 3subst Foo of Str7⏏5 where *.chars > 10; say Foo.^refinee␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement…»
jnthn m: subset Foo of Str where *.chars > 10; say Foo.^refinee
camelia rakudo-moar 6cd6ef: OUTPUT«(Str)␤»
jnthn m: subset Foo of Str where *.chars > 10; say Foo.^refinement
camelia rakudo-moar 6cd6ef: OUTPUT«{ ... }␤»
El_Che jnthn: this is a very naive implementation of wrapper like you said using EVALFILE: paste.ubuntu.com/23061364/ . My question: is there something like EVALFILE that would implement the perl6 -c behaviour? Some magic with the $lang option?
jnthn m: subset Foo of Str where *.chars > 10; say Foo.^refinement('abc') 12:19
camelia rakudo-moar 6cd6ef: OUTPUT«Too many positionals passed; expected 2 arguments but got 3␤ in any refinement at gen/moar/m-Metamodel.nqp line 3441␤ in block <unit> at <tmp> line 1␤␤»
jnthn m: subset Foo of Str where *.chars > 10; say Foo.^refinement().('abc')
camelia rakudo-moar 6cd6ef: OUTPUT«False␤»
gfldex ENOSUCHDOC
gfldex takes notes
jnthn El_Che: Not yet, but if use "use nqp;" then you can nqp::getcomp('perl6') and it's got methods on to let you do that... 12:20
(Bit tied up with $dayjob at the moment to find an example, sorry)
El_Che jnthn: thx. No haste
El_Che jnthn: the question is if I continue trying to parse the error string and learn more vimscript of try to get into nqp :) 12:21
I can't win
lizmat jnthn: isn't the code in sub EVAL (except the last line) really what we want ? 12:22
El_Che (I thought that adding perl6 support to syntastic was't nothing more than copying the existing perl5 one with a few adaptations. Famous last words. At least it kind of work already, but as I say, too fragile)
CIAvash unmatched}: The reddit comments didn't make me discuss the errors, if that's what you're implying. I was thinking about it before even the Rust article. 12:23
andreoss can i subset anonymous role/class? 12:32
gfldex m: subset Foo of role :: {};
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Invalid typename 'role'␤at <tmp>:1␤------> 3subset Foo of role7⏏5 :: {};␤»
gfldex m: subset Foo of class :: {};
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Invalid typename 'class'␤at <tmp>:1␤------> 3subset Foo of class7⏏5 :: {};␤»
unmatched} CIAvash: I guess it's not just the fallout from the article that bugs me, but a ton of LTA error tickets in RT queue. They all feel disjointed, with no master plan for them. Last week I changed a content of a warning, to close a ticket, only to realize it was changed earlier not so long ago, also due to "LTA error" ticket. And today's discussion technically means that warning is LTA again and needs to be 12:33
gfldex nope
unmatched} changed again. So we just keep changing stuff about any time someone has an alternate view on what the text should read.
nine El_Che: what lizmat said :)
El_Che: have a look at src/core/control.pm
lizmat nine: wonder whether we should expose this as COMPILE() 12:34
unmatched} CIAvash: and maybe there should be (or maybe there already is) a mechanism for a module-space module to take all these exceptions these errors actually are and output them and format them any way the module pleases. We can have Exceptions::Rust Exceptions::Elm Exceptions::Trans::Russian, whatever
*shrug*
El_Che nine, lizmat: having a look 12:35
nine, lizmat: although alternative structured (like yaml or json) error would also be cool: perl6 -c --errors-as-json :) 12:36
nine: suprisingly, the crappy github functionality gave me the correct file as the second match. It never does that 12:37
most of the time I end up cloning and using ack :)
dalek osystem: c924372 | (Alexey Melezhik)++ | META.list:
Sparrowdo module to install Nginx web server.
 12:38
osystem: 43c634f | (Zoffix Znet)++ | META.list:
Merge pull request #233 from melezhik/master

Sparrowdo module to install Nginx web server.
CIAvash unmatched}: I understand what you mean, but I don't think just dicussing it would hurt anybody. I'm not saying we have to do something right now, or do it based on someone's comments on reddit. And you can just share your opinion like you did, discussions shouldn't necessarily lead to some action. 12:41
arnsholt El_Che: The Perl 6 IPython kernel has an example of getcomp and friends, just gimme a sec to find it 12:42
andreoss gfldex: is it bug/NYI?
unmatched} andreoss: what's the usecase for this? If you're "throwing" the role away anyway, why not make it be right from the start, without a subset? 12:44
arnsholt El_Che: github.com/timo/iperl6kernel/blob/...6.pm6#L177 is the code that evaluates incoming code
gfldex andreoss: i don't think it makes sense in the first place. But then, I do think a lot of odd things.
arnsholt It's a bit complex due to having to thread the context of the previously evaluated code around the new code
andreoss unmatched}: to add attributes to subtype. `subset Foo of role :: { does Bar; has $.zzz }; ` 12:45
jnthn A subtype doesn't have state
arnsholt There are two places you want to look at for examples of how Rakudo does it today: src/Perl6/Compiler.nqp in the rakudo repo, and src/HLL/Compiler.nqp in nqp
jnthn *subset type
unmatched} andreoss: so just do role Foo does Bar { has $.zzz }
jnthn So no, it doesn't make sense.
andreoss paste.debian.net/789857/ 12:47
unmatched}: i was trying to use subsets as ADTs
which now i see is mis-use 12:48
gfldex m: subset Nil of Str; 12:53
camelia ( no output )
gfldex o.0
El_Che arnsholt: thx. But I think you code really executes the code (that would imply getting -I right), while I think for I linter I just need to compile it. So back to the last line of EVAL like liz said.
unmatched} m: class Nil {}; role Str {}; 12:54
camelia ( no output )
gfldex that takes obfuscation to a completely now level :) 12:55
arnsholt El_Che: Yeah, for that there are different methods on the compiler object you can call. I *think* the one you want is simply called compile 12:57
Which returns a callable object which can be invoked to do the execution
unmatched} m: sub say { put "You shall not say! $^a" }; sub infix:<+> { $^a - $^b }; say 2 + 2 12:59
camelia rakudo-moar 6cd6ef: OUTPUT«You shall not say! 0␤»
arnsholt I should probably write up some kind of description of how to get at and interact with the compiler object; it's less horrifyingly guts hackery than you might expect 13:02
In my ample spare time, in between thesis, new job and baby >.< 13:03
El_Che arnsholt: yeah, join the club
arnsholt: that should go right in to the doc (next to the blog). To useful to get lost in a random post 13:04
andreoss m: class Str {}; say "hi"; 13:05
camelia rakudo-moar 6cd6ef: OUTPUT«===SORRY!===␤This type cannot box a native string: P6opaque, Str␤»
andreoss why no "Redeclaration of symbol Str"? 13:06
unmatched} You can shadow all of the builtins by default.
*by design
jnthn Because the existing one is a lexical symbol
And so the new package-scoped decl doesn't conflict
arnsholt The setting is an outer lexical scope, so you get shadowing instead 13:07
jnthn Right
arnsholt It's like "my $a = 1; { my $a = 2; }" 13:07
andreoss m: our class Bar {} ; { local class Bar {} };
camelia rakudo-moar 6cd6ef: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Redeclaration of symbol Bar␤at <tmp>:1␤------> 3our class Bar {} ; { local class Bar7⏏5 {} };␤ expecting any of:␤ generic role␤»
andreoss m: my class Bar {} ; { my class Bar {} }; 13:08
camelia ( no output )
jnthn Closer to "my $a = 1; { our $a = 2; }" I guess, but yeah. 13:09
unmatched} m: my class Bar {} ; { class Bar {} };
camelia ( no output )
pmurias unmatched}: re comments on reddit, most of them seemed people nit picking for arguments sake 13:13
SmokeMachine____ m: say Str(Int) ~~ Int 13:15
camelia rakudo-moar 6cd6ef: OUTPUT«False␤»
SmokeMachine____ Shouldn't it be true? 13:16
If it shouldn't, how can I get the Str and the Int from the Str(Int)? 13:17
SmokeMachine____ Something like: Str(Int).final-type == Str && Str(Int).from-type == Int 13:19
Is there something like that?
RabidGravy er 13:28
unmatched} m: my $v = Str(Int); say $v.^methods 13:30
camelia rakudo-moar 6cd6ef: OUTPUT«Method 'methods' not found for invocant of class 'Perl6::Metamodel::CoercionHOW'␤ in block <unit> at <tmp> line 1␤␤»
RabidGravy It's not a type as such, though it looks weird - it's a type coercion of a type object, I'd expect a different result 13:31
unmatched} m: my $v = Str(Int); say $v.^target_type
camelia rakudo-moar 6cd6ef: OUTPUT«(Str)␤»
unmatched} m: my $v = Str(Int); say $v.^constraint_type
camelia rakudo-moar 6cd6ef: OUTPUT«(Int)␤»
RabidGravy yeah
my brane wants Str(Int) to be an undefined string 13:32
gfldex Perl6::* are missing in the docs, are they in roast? 13:35
gfldex seams not so 13:35
RabidGravy gfldex, they are documented as MetaModel::* e.g. docs.perl6.org/type/Metamodel$COLO...eContainer 13:36
gfldex a few are 13:36
RabidGravy so CoercionHOW isn't 13:37
well it's add the ones that are tested and not documented
gregf_ m: my $v = Str(Int); say $v.^name 13:43
camelia rakudo-moar 6cd6ef: OUTPUT«Str(Int)␤»
gregf_ m: my $v = Str(Int); say $v.^name; $v = 10.29; 13:44
camelia rakudo-moar 6cd6ef: OUTPUT«Str(Int)␤»
TheLemonMan hmm, the exception backtrace from a broken promise points to the block where await is called, quite counterintuitive when you have to debug some code gone wild 13:44
gfldex known bug 13:45
unmatched} Is it really a bug, rather than merely a limitation?
gfldex you can add a CATCH inside the thread and .say the exception
unmatched} TheLemonMan: it's cause it's on another thread. Stick a CATCH { default { warn $_; warn .backtrace} }
somewhere inside that thread 13:46
TheLemonMan that's much better 13:46
unmatched} \o/
Definitely an area of improvement to be made in. Gets pretty annoying when you got lots of starts and supplies and things 13:47
TheLemonMan you'd have to somehow raise the exception (or pass some more informations about the call-site) rather than just passing the Exception object as a result 13:48
gfldex IIRC it's in a Failure already but await is messing it up 13:49
TheLemonMan afaics await is just calling .result on the promise, which in turn just calls .throw 13:51
RabidGravy I think if you catch the exception from the await and get the inner exception or something it's right 13:51
ugexe you have to catch the 3rd exception, but only during a blood moon 13:52
unmatched} :D
TheLemonMan hmm, will a full moon suffice ? 13:56
ugexe are you willing to part with your first born?
RabidGravy speaking of blood sacrifice I'd better be testing the Net::AMQP with this new rabbitmq version 13:58
dalek osystem: 51b7ca0 | (Alexey Melezhik)++ | META.list:
Sparrowdo module to install/update Sparrow Toolchain on target host.
 14:14
osystem: 4376a98 | (Zoffix Znet)++ | META.list:
Merge pull request #234 from melezhik/master

Sparrowdo module to install/update Sparrow Toolchain on target host: github.com/melezhik/sparrowdo-sparrow-update/
SmokeMachine____ unmatched}: thanks! 14:16
pmurias new nqp-js blog post: blogs.perl.org/users/pawel_murias/2...pdate.html 14:24
unmatched} pmurias++ 14:27
CIAvash pmurias++ 14:29
mst awesome! 14:38
defend perl6's naming on twitter, get yelled at by stmuk 14:39
unmatched} ooo drama :D 14:39
mst: link? :)
mst twitter.com/shadowcat_mst/status/7...3764837377 14:40
RabidGravy moar popcorn needed, between that and people roasting UKIP :) 14:42
unmatched} Ah, great, ribasushi calling my words schizzophrenic. 14:43
unmatched} sighs
RabidGravy there are some particularly uncivilised people around
gfldex twitter could be so nice without all those humans
mst unmatched}: I thought my defence was really quite nice 14:44
until for some reason the star release manager turned up and started yelling at me and the perl6 person in the thread :(
gregf_ heh.. cool down guys.. its Perl *runs*
El_Che ah, I am famous. I am in the twitter thread (nxadm) 14:45
unmatched} mst: I don't see those comments though for some reason
mst twitter.com/nxadm/status/765545028026527745 14:46
El_Che unmatched}: it starts here twitter.com/ribasushi/status/765500776034758656
unmatched} heh 14:47
RabidGravy everyone will be famous for some multiple of 160 characters 14:47
El_Che anyway, there seems to be some bad blood there. Not much of a discussion after a few tweets 14:48
mst I was trying to constructively point out that the will to rename wasn't even -really- there on the perl5 side 14:49
given my plan got basically ignored
mst also, 'larry gets to name things perl' <- uh, yeah, that's why I was hoping we'd end up with 'pumpkin perl' and 'camelia perl' as being what people said 14:49
El_Che same same but different, not cool - let's be friends - I tried to fix it but nobody moved - shut up - shut up you first
there you go :)
mst right 14:50
still confused why smtuk decided to jump in and start attacking people
but, whatever
jdv79 i wonder if that squabble will ever die 14:52
mst I like my answer of "I tried, people didn't, let's file it under 'actually people don't care that much'" 14:53
unmatched} No, of course not. People don't really give a shit about the name, they just want their 5 minutes of fame yelling about things.
They could call it Rakudo, but no one cares enough.
pmurias I hope the better Perl 6 becomes the better the relations with the Perl 5 community will be
El_Che while the battle are the Field of Despair rages for days, people outside the echochamber thought: WUT, people still use PERL? :) 14:55
mst indeed
this is why my current project is building something non-perl people will want to use that's incidentally implemented in perl
it will probably eventually be implemented in a mixture of perl5 and perl6
but I figure the best way to advocate a language is to have people want to learn it because they're already using it for something 14:56
mst there's plenty of people who only know PHP because they needed to beat wordpress or mediawiki with a stick etc. 14:56
El_Che mst: that why I learn Ruby (Puppet)
t
unmatched} pmurias: if it weren't for mst, my relations would consist of (1) "Mojolicious developers shitlisting me for an exasperated tweet about their work and having a discussion about me for several hours" (2) Ribasushi parading my comments about wanting to live in peace as "schizzo". I can totally see this ending up in a breaking point and me stopping to care what Perl 5 folks think or do, since my only point of 14:57
contact is through these sort of interactions.
mst what I'm upset about is that I went to the trouble of stepping in to defend from the perl5 side
and then got yelled at by a perl6 person
it makes me not want to bother
El_Che mst in an ambassador role. Who would have thought :)
unmatched} :) 14:58
lizmat El_Che: you might be surprised :-)
El_Che I ... am
:)
mst El_Che: ironically, the only reason Chef was written in ruby was that their early customers were all rails shops
the prototype was a Catalyst app
El_Che mst: I was told that the first prototypes/versions of puppet were in perl, but the author moved to ruby for better dsl support 15:00
I can't find a verification of it though 15:01
mst El_Che: yes, that's *puppet*. I said *chef*.
I was around for the birth of both.
El_Che that sounds creepy :) 15:02
mst the ghost of christmas past
timotimo i don't see stmuk be a part of that discussion at all, strangely 15:18
mst starts from twitter.com/nxadm/status/765545028026527745 with him attacking El_Che for no apparent reason
timotimo oh, man. twitter is hard to navigate. 15:19
*shrug* :\ 15:20
gregf_ imho "radically different languages" doesn't sound quite right :| 15:22
gfldex m: use variables :D; my @a; @a[0] = Any; 15:23
camelia rakudo-moar c159f0: OUTPUT«Type check failed in assignment to @a; expected Any:D but got Any (Any)␤ in block <unit> at <tmp> line 1␤␤»
stmuk El_Che: there is a FLOSS Weekly interview with the Puppet authors where they say they were Perl people who used Ruby for pragmatic reasons 15:41
mspo I recall it 15:42
back in the great ruby DSL trend days
timotimo the great DSL wars
mspo and puppet is dog slow to prove how ruby-esque it is 15:43
my puppet runs are 30s - 3m now-a-days
it's painful
unmatched} What's an R operator? Trying to search for it on docs site is proving impossible :/ 15:44
timotimo it's a metaoperator
it reverses the arguments to an operator
TheLemonMan m: my @y is default(42); say @y;
camelia rakudo-moar c159f0: OUTPUT«[]␤»
timotimo m: say 5 - 10; say 5 R- 10
camelia rakudo-moar c159f0: OUTPUT«-5␤5␤»
timotimo m: say 5 / 10; say 5 R/ 10
camelia rakudo-moar c159f0: OUTPUT«0.5␤2␤»
unmatched} Oh
m: say 5 R, 1
camelia rakudo-moar c159f0: OUTPUT«(1 5)␤»
unmatched} Thanks
timotimo m: 10 R= my $foo; say $foo 15:45
camelia rakudo-moar c159f0: OUTPUT«10␤»
unmatched} 0.o
Ulti SmokeMachine____ your example with the scalar and hash version of SetHash the % named variables are all Hashes so if you say SetHash %variable you end up declaring a Hash[SetHash] where the keys are typed iirc
El_Che mspo: I restricted my puppet use to a run in a tmp space on a remote server to create the persistent configuration of docker containers (using among other hiera-eyaml for gpg encryption of secrets)
unmatched} m: my @y is default(42); say @y[^3]; # TheLemonMan, the default applies to elements that don't have values
camelia rakudo-moar c159f0: OUTPUT«(42 42 42)␤»
timotimo Ulti: nope, for typed keys you need to Hash{SetHash}
Ulti timotimo its more it ends up being that 15:46
not saying the square bracket thing is syntax
timotimo Ulti: the role arguments for Associative and Hash are "reversed" for that reason
Ulti just a description
El_Che mspo: so the puppet/ruby lib code is nowadays pretty simple and small (it used to be huged and complicated)
TheLemonMan unmatched}, ugh, that's confusing
Ulti %thing is a hash I noticed this when someone else was asking about the exact same thing with BagHash
timotimo but they want a hash that has BagHash-es stored in it 15:47
Ulti no they want the %thing to be the BagHash at least in the person previously
unmatched} TheLemonMan: what did you think it would end up as?
gfldex m: my %h is SetHash.=new; dd %h.push(a => 1)
camelia rakudo-moar c159f0: OUTPUT«Cannot modify an immutable SetHash␤ in block <unit> at <tmp> line 1␤␤»
Ulti my BagHash %thing = BagHash.new;
unmatched} m: my $y is default(42); say $y; 15:48
camelia rakudo-moar c159f0: OUTPUT«42␤»
Ulti gfldex: hmm interesting
unmatched} m: my $y is default(42); $y = (1, 2, 3, 4,); say $y;
camelia rakudo-moar c159f0: OUTPUT«(1 2 3 4)␤»
TheLemonMan unmatched}, a compile-time error tbh heh, I originally mistyped the sigil in my code and was wondering what was going on
Ulti so the `is` is needed
timotimo Ulti: that's just a bug; i think it was already filed.
Ulti k k
SmokeMachine____ Util: no iPhone wanted a hash of SetHashes... 15:49
[Coke] would give lizmat a hug, but is afraid that would hurt. 15:51
SmokeMachine____ My problem was I couldn't set values on a sethash inside of a hash... 15:52
lizmat [Coke]: there are still spots unbruised :-)
El_Che lizmat: what happened? 15:52
lizmat last Sunday hit the concrete of the bikepath at 25km/hour 15:53
SmokeMachine____ m: my SetHash %a; %a<bla><ble>++
camelia rakudo-moar c159f0: OUTPUT«Type check failed in assignment to %a; expected SetHash but got Hash (${})␤ in block <unit> at <tmp> line 1␤␤»
El_Che lizmat: let me guess, you wanted to speed up stuff with at least 30%
:)
lizmat: something broken?
lizmat nothing broken, just a lot of scratches and bruises
I now have green fingers on my left hand :-) 15:54
quite amazing :-)
AlexDaniel lizmat: what about the bike? Is it alright?
lizmat yeah, mostly, one of the brakes needed adjusting 15:54
[Coke] ROTFLS at el_che's comment. :)
SmokeMachine____ Shouldn't that work? 15:57
AlexDaniel lizmat: well, consider rethinking your priorities :D that being said, some people argue that while healhcare is free in some countries, bike parts are not…
unmatched} lol 15:58
lizmat AlexDaniel: and don't forget the labour :-)
gfldex SmokeMachine____: no, it should not work 15:59
SmokeMachine____ Why not?
gfldex because you never told the compiler the shape of %a. Therefor it must assume that you want a "autovivified" shape. 16:00
and shaped Hashes are NYI
(i think)
m: my SetHash %a{1}; 16:01
camelia ( no output )
gfldex m: my SetHash %a{1}; dd %a
camelia rakudo-moar c159f0: OUTPUT«Hash[SetHash,Int] %a = (my SetHash %{Int} = )␤»
dalek c: 8105200 | coke++ | util/list-missing-methods.p6:
whitespace cleanup
 16:02
c: 6b162d4 | coke++ | doc/Type/Callable.pod6:
whitespace
SmokeMachine____ I said that I wanted a hash of sethashes... Didn't I? 16:03
gfldex m: my %a; %a<bla> = SetHash.new; %a<bla><blub>++; dd %a; 16:03
camelia rakudo-moar c159f0: OUTPUT«Hash %a = {:bla(SetHash.new("blub"))}␤»
SmokeMachine____ But why it's not autovivified? 16:04
unmatched} m: my SetHash %a; %a<bla> = SetHash.new; %a<bla><ble>++
camelia ( no output )
unmatched} m: my SetHash %a; %a<bla> = SetHash.new; %a<bla><ble>++; dd %a
camelia rakudo-moar c159f0: OUTPUT«Hash[SetHash] %a = (my SetHash % = :bla(SetHash.new("ble")))␤»
unmatched} SmokeMachine____: it is; it's just the autovivification uses a Hash, so you get the type mismatch error 16:05
gfldex m: my %a is default(SetHash.new); # it should do that but with a proper closure
camelia ( no output )
Ledmitz hello. this should be a simple one sine I don't know perl at all. having a problem with quoted text. I think the "s/" seems to be the problem. How can I treat as normal quote?
"/media/Silo/Projects/Audio/ManaSound/Done"
unmatched} Ledmitz: there's no problem in that string. We'll need more context. 16:06
Ledmitz: also, what Perl is that? Perl 5 or Perl 6?
SmokeMachine____ m: my SetHash %a; %a<bla> .= new; %a<bla>++; dd %a
camelia rakudo-moar c159f0: OUTPUT«Method 'succ' not found for invocant of class 'SetHash'␤ in block <unit> at <tmp> line 1␤␤»
Ledmitz ok .. its the bfserve.pl script. I want to host sfx that I have made 16:06
unmatched} Ledmitz: are there error messages? 16:07
SmokeMachine____ But if I said that it's a sethash, shy it autovivifies as a hash? 16:07
Ledmitz OOOhhhh unmatched} ... I am Perl 5 here, I guess
gfldex SmokeMachine____: you may have stepped on a bug 16:08
Ledmitz unmatched}> shall I send a code snippet to you pm?
unmatched} Ledmitz: OK.
Ledmitz: and error messages you're getting
SmokeMachine____ gfldex: already known bug? 16:09
gfldex _may_
Ledmitz unmatched}> no errors and I checked syntax online. all good, but terminal shows different color 16:09
gfldex the problem is that the type you provide, may require arguments to new to make sense. How should the compiler know? 16:10
autovivication with %h<foo>++ makes sense because ++ defaults to Int and Int.new defaults to 0 so you end with a foo => 1 16:11
unmatched} Ledmitz: erm, you're just using a dumb syntax highlighter that thinks it's the s/// operator, even though it's just a string.
Ignore it.
SmokeMachine____ Shouldn't it looks at the type of the variable to get a hint? 16:12
gfldex it doesn't get there for %a<bla><ble>++ 16:13
Ledmitz unmatched}> ok.. just trying to track down why the server not functioning.
unmatched} Ledmitz: you might get better luck in one of... I'm assuming Battle Field... forms, since the problem is likely not with the script, but with your setup. 16:14
*forums
gfldex m: my SetHash %a; %a.VAR.default.say
camelia rakudo-moar 05cdc1: OUTPUT«(SetHash)␤»
gfldex so the default value is a type object
m: my SetHash %a is default(Any); %a.VAR.default.say 16:15
camelia rakudo-moar 05cdc1: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Default value '(Any)' will never bind to a parameter of type Hash[SetHash]␤at <tmp>:1␤------> 3my SetHash %a is default(Any)7⏏5; %a.VAR.default.say␤ expecting any of:␤ constraint␤»
Ledmitz unmatched}> hm.... I'm guessing the script is really old and the pearl newest on Debian Unstable
SmokeMachine____ m: my SetHash %a; %a<bla>.^name.say
camelia rakudo-moar 05cdc1: OUTPUT«SetHash␤»
Ledmitz *new
unmatched} Ledmitz: it should still work fine. Maybe your config is wrong, dunno. 16:16
Ledmitz unmatched}> ok well ty. where do I go for forums?
SmokeMachine____ Shouldn't it give a hint?
unmatched} Ledmitz: no idea. This is a Perl 6 programming channel and your script is a Perl 5 server for a game :} Not exactly overlapping fields :)
Ledmitz unmatched}> well sorry bout that. I'm gonna try #hexchat. they might know. could be client settings 16:17
SmokeMachine____ m: my SetHash %a is default(SetHash.new); %a<bla><ble>++; dd %a
camelia rakudo-moar 05cdc1: OUTPUT«Hash[SetHash] %a = (my SetHash %)␤»
Ledmitz unmatched}> channel modes seem fine
Ledmitz thanks 16:18
unmatched} No problem.
gfldex is default(SetHash.new) will call .new at compile time 16:19
SmokeMachine____ Makes sense...
gfldex so you end up with the same Sethash in any slot of %a
gfldex m: my SetHash %h; %h but WHENCE({SetHash.new}) 16:22
camelia rakudo-moar 05cdc1: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Undeclared name:␤ WHENCE used at line 1␤␤»
gfldex is WHENCE implemented at all? 16:25
nvm, it's not in roast 16:26
lizmat gfldex: WHENCE *is* implemented at nqp level and is responsible for:
my %h; my $b := %h<a>; dd %h; $b = 42; dd %h 16:27
m: my %h; my $b := %h<a>; dd %h; $b = 42; dd %h
camelia rakudo-moar 05cdc1: OUTPUT«Hash %h = {}␤Hash %h = {:a(42)}␤»
lizmat afk&
timotimo oh wow, this is comedy gold 16:31
timotimo "it is literally the same meaning as its dictionary definition" - "no it's not, *links to oatmeal comic about 'literally'*" - "yeah, it is." 16:31
unmatched} is confused 16:32
timotimo the "literally" was properly used in the original comment
and someone went "LITERALLY ISNT USED LIKE THAT"
SmokeMachine____ m: my %a is Hash of SetHash; %a<bla><ble>++; dd %a 16:33
camelia rakudo-moar 05cdc1: OUTPUT«(my SetHash % = :bla(${:ble(1)}))␤»
timotimo though of course you can debate whether a function "taking" a code block is an acceptable use of a word that literally means "taking"
unmatched} feels like the context is missing...
timotimo just the comment thread about your blog post 16:34
SmokeMachine____ gdldex: looks that it worked...
unmatched} Ah
timotimo i wonder if clang had levenshtein based suggestions before rakudo did
unmatched} I like gobbling. It's cute and fun. An error in your code is a traumatic event, so humour is a good thing.
SmokeMachine____ Or not... 16:35
m: my %a is Hash of SetHash; %a<bla><ble> = "qwer"; dd %a
camelia rakudo-moar 05cdc1: OUTPUT«(my SetHash % = :bla(${:ble("qwer")}))␤»
timotimo i think that's a bug there 16:36
probably already filed
*shrug*
apparently clang already had typo correction in 2011 at the latest 16:38
unmatched} Lool, 44 comments on that reddit thread. "Did you actually read the error?" "Did you actually read my comment?".. Yeah, I don't think I going to be reading them :} 16:39
unmatched} contemplates publishing "Perl 6 The Most Awesome Language You Ever Dreamed Of!" 16:41
And then watch the world burn, as people argue themselves in the comments to death :D
I might do it just to troll...
timotimo not sure how i'd like that 16:44
stmuk social media would be fine if it wasn't for the people that used it!
unmatched} timotimo: but it would be purely for my own entertainment :D 16:47
timotimo OK 16:48
unmatched} m: say (1 R, 2 R, 3 R, 4) 16:51
camelia rakudo-moar 05cdc1: OUTPUT«(2 1)␤»
mspo El_Che: so you just use it as a Dockerfile template machine? 16:52
[Coke] unmatched}: precedence 16:54
unmatched} [Coke]: of what? I can't figure out how to put parens in 16:55
m: say (1 R, (2 R, (3 R, 4)))
camelia rakudo-moar 05cdc1: OUTPUT«(((4 3) 2) 1)␤»
unmatched} Ah
hm 16:56
m: say (((1 R, 2) R, 3) R, 4)
camelia rakudo-moar 05cdc1: OUTPUT«(4 (3 (2 1)))␤»
unmatched} m: @a = (1 R, 2 R, 3 R, 4); dd @a
camelia rakudo-moar 05cdc1: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Variable '@a' is not declared␤at <tmp>:1␤------> 3<BOL>7⏏5@a = (1 R, 2 R, 3 R, 4); dd @a␤»
unmatched} m: my @a = (1 R, 2 R, 3 R, 4); dd @a
camelia rakudo-moar 05cdc1: OUTPUT«Array @a = [2, 1]␤»
unmatched} Come come (1, 2, 3, 4) doesn't have this issue? 16:57
[Coke] m: say |(1 R, 2 ) R, |(3 R, 4)
camelia rakudo-moar 05cdc1: OUTPUT«(4 3 2 1)␤»
timotimo so the R metaop doesn't do well with list prefix ops? 16:58
unmatched} It's from this ticket: rt.perl.org/Ticket/Display.html?id...et-history 16:59
[Coke] unmatched}: my plan on that ticket was to hope jnthn answered it. :| 17:00
unmatched} :D 17:01
jnthn I suspect that it just ain't handling list associativity right 17:09
jnthn@lviv:~/dev/rakudo$ ./perl6-m --target=ast
===SORRY!===
Cannot find method 'repl-loop' on object of type QAST::CompUnit
Well, bother. 17:10
Curiously though the AST looks kinda right 17:11
so does METAOP_REVERSE 17:12
Oh
The optimizer screws it
With --optimzie=off it works out fine
timotimo uh oh. i wonder if i tried to "optimize" the metaop reverse? 17:13
jnthn Probably :P 17:14
Compiling a patch
timotimo needs moar tests 17:15
jnthn yup, got a fix 17:16
unmatched} \o/ 17:16
jnthn Of course, my NQP and Rakudo builds are way behind HEAD... :) 17:19
m: use Test; is (1 R, 2 R, 3 R, 4), (4, 3, 2, 1) 17:21
camelia rakudo-moar 05cdc1: OUTPUT«not ok 1 - ␤␤# Failed test at <tmp> line 1␤# expected: '4 3 2 1'␤# got: '2 1'␤»
unmatched} m: use Test; is-deeply (1 R, 2 R, 3 R, 4), (4, 3, 2, 1) 17:22
camelia rakudo-moar 05cdc1: OUTPUT«not ok 1 - ␤␤# Failed test at <tmp> line 1␤# expected: $(4, 3, 2, 1)␤# got: $(2, 1)␤»
El_Che mspo: I am not a linux admin jobwise, so that part is on their team (jenkins, git setup etc). For our team I picked the pieces togethere and wrote what was needed to allow us to do our deployments on a safe way. We ended with a server that retrieve the configuration from git, build it, puts the secrets in (in git encrypted with gpg hiera-eyaml) and pushed the config to the docker host and restarts the 17:25
container (or retrieve the coorrect image from the registry)
mspo El_Che: oh okay 17:31
El_Che mspo: it works pretty nice, a git-jenkins flow 17:32
mspo: but becasue of security concerns we needed a more specific flow (we are responsable for the AAI setup (infra & code), including everything related to ldap, ad, iam, radius, etc) 17:34
kyclark If I “run” with :merge, I never get anything back. Minimal failing case:
pastie.org/10936630
jnthn unmatched}: Fix pushed, ticket closed :) 17:35
unmatched} kyclark: it's likely buggy. I get a coredump with my $p = run "echo", "hello", :merge; say $p.out.slurp-rest 17:36
jnthn: that was fast :D \o/
jnthn Yeah, well, my alternative for the last 30 mins of work time today was continuing dealing with integrating with an oAuth2 impl whose authors can't read RFCs...so it wasn't really a hard choice. :P 17:37
unmatched} heh
stmuk remembers the YAPC::EU 14 oauth dance 17:59
brrt o... auth... :-( 18:07
although, i was actually having problems with xauth today
so
always trouble
El_Che jnthn: come to saml2 if you want an overdose of RFCs and xml :) 18:13
kyclark Is there a way to do more than one file test at a time. $file.IO.fe won’t give me “file” and “exists” 18:51
sjn kyclark: doesn't "file" imply "exists"? 18:53
kyclark Yes, bad example, what about “file” and “executable”? 18:53
I guess just $file.IO.e && $file.IO.f?
unmatched} kyclark: $file.IO ~~ :e & :f
kyclark Oooo, pretty~
!
unmatched} .e and .f given $file.IO 18:54
ugexe you used to be able to do .all<f e> 18:54
unmatched} kyclark: sorry, executable woulbe :x not :e 18:55
lizmat . 18:56
sjn btw, is there a reason for panda seemingly throwing an error when trying to install a module that's already installed? 18:57
ugexe thats actually rakudo that throws
kyclark Is it possible to set an INPUT_RECORD_SEPARATOR-somethingorother when reading files? 18:58
e.g., $/ 18:59
ugexe nl-in and nl-out
smls kyclark: Pass the :nl ("newline") option to open()
dLeCamarae Has anyone built Rakudo for the Intel Galileo, or will I be the first?
sjn ugexe: stack trace refers to Panda::Installer line 61 <github.com/tadzik/panda/blob/maste...pm#L61> 19:00
ugexe $to.install is calling a method on a rakudo class
sjn CompUnit::Repository
ugexe ::Installation 19:01
TheLemonMan I dug a bit into the ticket #128803 because apparently it's how I now like wasting my time, found that what goes wrong is is-runtime doesn't return True when called in print-exception, I instrumented with some say the code and this is the output I got from running the first test-case outlined in the ticket ptpb.pw/jpWv I hope that's somehow useful :) 19:31
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128803
unmatched} TheLemonMan: great! Can't wait for your fix :) 19:38
unmatched} m: X::OutOfRange.new(what => ".rotor position is", got => -1, range => "0..Inf", comment => "(ensure the negative gap is not larger than the length of the sublist)").new.throw 19:46
camelia rakudo-moar ba20d3: OUTPUT«Argument out of range. Is: <unknown>, should be in <unknown>␤ in block <unit> at <tmp> line 1␤␤»
unmatched} Weird that the got/range/comment don't get displayed :\
unmatched} s: <a b c d b>, "rotor", \(1 => -2) 19:47
SourceBaby unmatched}, Sauce is at github.com/rakudo/rakudo/blob/1728...t.pm#L1152
unmatched} Ah, two .news
m: X::OutOfRange.new(what => ".rotor position is", got => -1, range => "0..Inf", comment => "(ensure the negative gap is not larger than the length of the sublist)").throw
camelia rakudo-moar ba20d3: OUTPUT«.rotor position is out of range. Is: -1, should be in 0..Inf; (ensure the negative gap is not larger than the length of the sublist)␤ in block <unit> at <tmp> line 1␤␤»
TheLemonMan unmatched}, heh I'm still trying to figure out how all the moving pieces work together 19:50
unmatched} TheLemonMan: if it's of help, there's an Internals course: edumentab.github.io/rakudo-and-nqp-...ls-course/ 19:51
TheLemonMan oh, that's very nice, thanks unmatched}
unmatched} There's also this article; unsure of how useful it will be to you tho: perl6.party/post/Hacking-on-Rakudo-...x-Your-Fix 19:52
TheLemonMan that's the article that made me try out Perl6 :) 19:53
[Coke] unmatched}++ 19:53
lizmat unmatched} : seems like my fix borked the test for #127695 in t/spec/S19-command-line/repl.t ? 19:57
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127695
unmatched} TheLemonMan: that's music to my ears :) 20:12
vcv unmatched}++
unmatched} lizmat, I'll take a look closer when I get home, but it seems to be just another test that assumed the REPL starts up even if the requested module was not found 20:14
lizmat unmatched}: that makes sense, safe travels!
TheLemonMan from what I can tell it seems the backtrace is too shallow to reach the point where eval/print_control/compile is 20:29
lizmat TheLemonMan: have you tried running with --ll-exception ? 20:39
TheLemonMan lizmat, yep, I've just learned about that switch heh 20:40
kyclark Given two arrays like @names and @values, how could I combine them into a single hash? 20:45
@names Z @values doesn’t quite do what I want
hoelzro @names Z=> @values 20:46
that should do the trick, I think
m: my @names = <foo bar baz>; my @values = 1..3; my %h = @names Z=> @values; say %h.perl
camelia rakudo-moar dd9b76: OUTPUT«{:bar(2), :baz(3), :foo(1)}␤»
kyclark Flippin’ sweet. 20:47
Heckin’ awesome.
perlpilot Perl 6 is just awesome that way :)
AlexDaniel interestingly, I'm no longer impressed by stuff like this 20:48
after seeing enough Perl 6 code it just feels normal 20:49
perlpilot This morning I was debugging some P5 and wanted to dump the value of a hashref on the 5th iteration of a loop. I immediately typed say Dumper($thingy) is $++ == 4; and then got annoyed that I had to use an actual variable and had to declare it somewhere.
s/is/if/ 20:50
perlpilot Perl 6 is full of little "convenience" things like that that just make it more fun to use 20:50
AlexDaniel it also annoys me that you have to import Dumper… 20:51
perlpilot indeed
El_Che m: class foo { foo() } 21:03
camelia rakudo-moar dd9b76: OUTPUT«WARNINGS for <tmp>:␤Useless use of constant value foo() in sink context (line 1)␤»
El_Che m: class foo { my $a = 'a'; foo } 21:04
camelia rakudo-moar dd9b76: OUTPUT«WARNINGS for <tmp>:␤Useless use of constant value foo in sink context (line 1)␤»
unmatched} that looks like a coercer
El_Che about the errors message. Regular errors (typo's, wrong infixes, forgotten ';', etc) return the filename in the first line of the error output. "Undeclared" errors (e.g. Undeclared routine) does not return a filename what so ever 21:06
zengargoyle it's been a while, but i think i've seen horrid SHA hashes in some sorts of error messages or other outputs where one would expect a 'filename' of some sort. 21:08
El_Che zengargoyle: it doen't matter that much as long as the errors are consistent. The filename (and line nr) is needed by the vim linter (syntastic) in order to highlight the error line 21:10
zengargoyle sure, i'm just thinking more of the general useage case. 21:11
geekosaur zengargoyle, that was fixed some time ago so it gets translated to the user visible name
perlpilot zengargoyle: sounds like you're thinking of how files used to be shown from comp unit repos 21:12
geekosaur (actually it outputs both, because if you have two different modules that use the same user visible names but are distinct by, say, auth or ver, you need the hash to distinguish)
zengargoyle cool, thought i saw one a week or two ago. but i haven't been doing much p6 for a while so ...
like i could have swore i saw it in like 'panda --help' or something where '/path/to/panda' was '/path/to/some-long-hash' or something. 21:14
perlpilot if you saw it a week or so ago, it must have been from an old perl6 :) 21:15
or an old panda maybe
unmatched} Don't you still get the hash in the filename, it's just you're also told what module it is in parentheses?
zengargoyle nah, fresh rakudobrew. install.
maybe i missed the parenthesized part. 21:16
perlpilot must be
(it's parenthetical ... how important could it be? ;)
unmatched} m: class foo { foo().HOW.^name.say }
camelia rakudo-moar dd9b76: OUTPUT«Perl6::Metamodel::CoercionHOW␤»
unmatched} \o/ 21:17
zengargoyle :)
geekosaur mm, this sounds like something different? like the script is printing its $*PROGRAM and that contains the hash because that's the actual name of the script within the filesystem?
(or is it $?PROGRAM? sigh)
unmatched} $*PROGRAM 21:19
zengargoyle ah, yeah. maybe it was some other older thing from pre precomp days that was printing out itself. 21:28
kyclark I have a MAIN that accepts a :$sep=‘,’ for parsing files that works fine with commas, but now I want to specify —sep=“\t” on the command line and it’s not working correctly to “split” the text. Any ideas? 21:30
ugexe are you sending an actual tab or a literal slash literal t? 21:31
kyclark I’m not sure how to specify it from the command line. I’ve tried —sep=\t —sep=“\t” —sep=\\t 21:33
None work
ugexe you have to do that yourself 21:34
kyclark Wow, an actual tab makes it funky
geekosaur suspects it only understands literal characters, not backslash escapes
you might want to try --sep=$'\t'
(bash-ism)
avuserow_ try `--sep="<Ctrl+v><Ctrl+i>"` (e.g. pushing ctrl+v then ctrl+i while within the double quotes)?
ugexe or s/\\t/\t/ in your program 21:35
unmatched} There's a site someone linked to here that makes you solve all sorts of decoding challenges in HTTP headers. Anyone recall the URL?
It was some different language, not English
kyclark geekosaur nailed it, but I guess I will have to detect literal ‘\t’ and set it to “\t” ??? 21:36
geekosaur what ugexe said, basically
geekosaur except you probably want general character escape processing (hm. is that perhaps exposed somewhere convenient?) 21:37
ugexe just use .trans, there isnt that many instances
kyclark Oh, right. ugexe got it, too 21:37
ugexe github.com/rakudo/rakudo/blob/nom/...ON.pm#L107 along those lines 21:38
geekosaur until you try to use \x or \c...
unmatched} Ah, found that site: www.vg.no/ 21:43
avuserow_ prepares for a flight by cloning `perl6-all-modules`... never know when you need some light reading 21:47
masak hm, perl6-all-modules hasn't been updated since May... 22:00
'night, #perl6 22:01
timotimo gnite masak 22:05
kyclark Can the arguments to MAIN be made mutable? 22:08
kyclark Here’s what I have right now: 22:09
sub MAIN (Str $file!, Str :$sep=',', Int :$limit=0) {
I need to alter $sep if it’s “\t” 22:10
geekosaur m: sub aMAIN (Str $file!, Str :$sep is rw =',', Int :$limit=0) { } 22:11
camelia rakudo-moar b00d92: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Cannot use 'is rw' on an optional parameter␤at <tmp>:1␤»
geekosaur bleh
kyclark Yeah, that’s what I was seeing. I wondered if I just had things in the wrong order.
geekosaur so, you can but you can;t use the =',' shorthand, you need to check for it being defined and set it to ',' yourself in code 22:12
kyclark It’s not a big deal to copy to another variable. Gives me an opportunity to discuss immutability.
geekosaur hm. no, tjat makes it mandatory. sigh
kyclark FWIW, I’m working up an intro to programming in Perl (6) for biologists (www.gitbook.com/book/kyclark/metag...s/details) 22:13
kyclark I don’t seem to be able to “run” a command with pipes e.g. “cut -f 1 | sort | uniq" 22:21
I can “shell” it, but it just goes to $*OUT
ugexe shell has the same options as run
El_Che are there perl 6 users using vim? I just pushed a fork of syntastic with perl6 support: github.com/nxadm/syntastic. It works for me (TM), but it still work in process (does the linter catch all the errors?) and need some work before sending a PR upstream. Perl6 info: github.com/nxadm/syntastic/blob/ma...E-perl6.md 22:23
beware: ugly vimscript code
geekosaur you can also make run spawn a shell to run the pipeline, although using shell is better. 22:24
ugexe you can pipe with run by doing like `run ..., :in($other-proc.out.lines)` 22:26
AlexDaniel geekosaur: what do you mean by “better” ? ;)
geekosaur I'm imagining that run splits its string into words. I doubt it does so with shell quoting, so you'd have to do various ugly things to keep it from producing ["/bin/sh", "-c", "'cut", "-f", "1", "|", ...] 22:27
El_Che This is wat it the perl6 support on vim-syntastic looks like: snag.gy/7LYGPV.jpg 22:28
geekosaur in fact you might not be able to prevent that at all... I haven't looked at how it does its thing
whereas shell doesn't need any string splitting hackery
kyclark OK, so I can run my pipeline with “shell” with a final redirect in to “> tmp” and then read that, but that’s fairly hackish, no? 22:32
AlexDaniel kyclark: .out.slurp-rest 22:35
run() does not split anything, you give it the arguments and they're passed as is 22:36
AlexDaniel shell() might not need string splitting hackery, but I don't know of any reliable way to pass any variables to it yet… you can pass your stuff into its STDIN, I guess… 22:38
so if you're choosing between splitting your arguments properly OR trying to prevent things to be evaled, I'd much rather choose the former 22:39
dalek ateverable: 26c5749 | (Daniel Green)++ | Bisectable.p6:
Show short commit hashes for 'good' commit also
 23:32
Ulti kyclark: in your book you might want to show given when with junctions rather than the if else for counting nucleotides 23:33
kyclark: also feel free to pinch anything from here perl6advent.wordpress.com/2014/12/...-of-perl6/ 23:35
timotimo m: constant FOOBAR == 4j # LTA error
camelia rakudo-moar b00d92: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Preceding context expects a term, but found infix = instead␤at <tmp>:1␤------> 3constant FOOBAR ==7⏏5 4j # LTA error␤»
timotimo the j was meant to be a ;, but oh well
j really looks a lot like a semicolon anyway
nxadm found that 23:38
El_Che timotimo: you type weird thing when testing the vim syntaxchecker :) 23:44
timotimo no, nxadm typed a weird thing when testing the vim syntaxchecker 23:46
El_Che I am nxadm 23:47
:)
and nxadm goed to bed, bye! 23:49
kyclark Ulti, thanks for all that! 23:53
Ulti, can you expand on your comment about junctions for counting nucleotides?
timotimo YOU ARE THAT! 23:54
MasterDuke kyclark: i think he means that: if $letter eq 'a' || $letter eq 'A' could instead be: if $letter eq 'a' | 'A' 23:58
timotimo yes, except the optimization for that is - if i'm not mistaken - tsill b0rked at the moment 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!