This channel is intended for people just starting with the Raku Programming Language (raku.org). Logs are available at irclogs.raku.org/raku-beginner/live.html
Set by lizmat on 8 June 2022.
01:10 MasterDuke left 01:37 MasterDuke joined 03:44 rf left 05:36 kjp left 05:39 kjp joined 06:31 Heptite left 06:42 Sussysham joined 06:43 Sussysham left 08:25 Sussysham joined 09:36 Sussysham left 10:48 Sussysham joined 10:51 Sussysham left 10:52 NemokoschKiwi joined 11:04 NemokoschKiwi left
el gatito (** advocate) m:perl say infix:<//>.WHY 12:23
Raku eval No documentation available for type 'Any'. Perhaps it can be found at docs.raku.org/type/Any
Nemokosch you missed &
el gatito (** advocate) m:perl say &infix:<//>.WHY 12:24
Raku eval No documentation available for type 'Sub+{Precedence}'. Perhaps it can be found at docs.raku.org/type/Sub+{Precedence}
Nemokosch welp, kinda sad
el gatito (** advocate) m:perl say &infix:<+>.WHY 12:25
Raku eval No documentation available for type 'Sub+{is-pure}+{Precedence}'. Perhaps it can be found at docs.raku.org/type/Sub+{is-pure}+{Precedence}
el gatito (** advocate) built in functions doesn’t have inspectable pod 😭 12:26
Nemokosch very much afraid this is for "performance reasons" 12:27
lizmat historically, yes
el gatito (** advocate) “we don’t put docs inside builtins because we are not python” 12:28
lizmat I don't think there was any thought like that ever 12:29
I do think we're in a position now to make it fetch documentation, or at least provide a better link nowadays 12:30
it's just low on the triage list
el gatito (** advocate) now there is no excuses not to 12:31
Nemokosch be ready for a "well volunteered", haha 12:32
lizmat ferry vunny 12:33
"pay it forward" I'd say 12:35
el gatito (** advocate) m: sub a {} #= italic bold say &a.WHY 12:38
Raku eval *italic* **bold** 12:39
el gatito (** advocate) m: sub a {} #= <C code> say &a.WHY
Raku eval <C code>
lizmat yes, that works *after* the settings have been compiled 12:40
it does *not* work at the moment the infix ops are compiled
bootstrapping is hard
el gatito Nemokosch github.com/rakudo/rakudo/commit/faff909125 13:05
el gatito (** advocate) how to pull out the invocant argument of the caller i have a &super function instead of having to write super(self) i want to just write super to call it on self 13:06
lizmat m: class A { method a() { b() } }; sub b() { dd CALLER::.hash<self> }; A.a # naive method 13:12
camelia Cannot access 'self' through CALLER, because it is not declared as dynamic
in sub b at <tmp> line 1
in method a at <tmp> line 1
in block <unit> at <tmp> line 1
lizmat m: class A { method a() { b() } }; sub b() { dd CALLER::.pairs.first(*.key eq "self").value }; A.a # sneaky, but still no go
camelia Rakudo::Internals::LoweredAwayLexical
lizmat in any case, using CALLER:: means you have no hope on any significant optimizations 13:13
in the (nearer) future, you could think of super() as a macro, and that would be able to do TRT 13:15
el gatito (** advocate) m:perl sub f { say CALLER } (sub { f })()
Raku eval (CALLER)
el gatito (** advocate) m:perl sub f { say CALLER } sub g { f } g
Raku eval (CALLER)
el gatito (** advocate) TRT? 13:16
lizmat The Right Thing
el gatito (** advocate) who cares 😎 13:17
lizmat docs.raku.org/language/packages.ht...o-packages 13:18
Nemokosch umm, I'm not sure the replies are carried through the IRC
just saying xd
the replied message, that is
lizmat you mean: [14:17:49] <discord-raku-bot><el gatito (** advocate)> who cares 😎
? 13:19
Nemokosch yep
what did it refer to?
lizmat I have no idea
Nemokosch well, the optimizations 13:20
el gatito (** advocate) > in any case, using CALLER:: means you have no hope on any significant optimizations
Nemokosch but yeah the point is that using discord replies can have surprising results on the IRC
lizmat well, if someone wants to use Raku in production, they *would* appreciate care being taken wrt to optimizations 13:21
yeah, perhaps we the bot would need to be improved, or the bridge should be closed down
el gatito (** advocate) m:perl sub f { dd CALLER } sub g { f } g;
Raku eval CALLER
el gatito (** advocate) ok nothing useful 13:22
m:perl sub f { dd CALLER; } sub g { $x = 1; f; } g;
Raku eval Exit code: 1 ===SORRY!=== Error while compiling /home/glot/main.raku Variable '$x' is not declared at /home/glot/main.raku:6 ------> ⏏$x = 1;
el gatito (** advocate) m:perl sub f { dd CALLER; } sub g { my $x = 1; f; } g;
Raku eval CALLER
el gatito (** advocate) m:perl sub f { dd CALLER::; } sub g { my $x = 1; f; } g; 13:23
Raku eval PseudoStash.new(("\$!" => Nil,"\$/" => Nil,"\$x" => Rakudo::Internals::LoweredAwayLexical,"\$¢" => Nil))
el gatito (** advocate) cent variable wtf
Nemokosch seems funky 13:24
el gatito (** advocate) m:perl sub f { dd CALLER::<$x>; } sub g { my $x = 1; f; } g;
Raku eval Exit code: 1 Cannot access '$x' through CALLER, because it is not declared as dynamic in sub f at main.raku line 2 in sub g at main.raku line 7 in block <unit> at main.raku line 10
el gatito (** advocate) 🤨 13:25
Nemokosch I think this is some deliberate protectional mechanism
rather than a technical constraint
lizmat indeed
el gatito (** advocate) m:perl sub f { dd CALLER::.hash<$x>; } sub g { my $x = 1; f; } g; 13:26
Raku eval Exit code: 1 Cannot access '$x' through CALLER, because it is not declared as dynamic in sub f at main.raku line 2 in sub g at main.raku line 7 in block <unit> at main.raku line 10
lizmat m: my $x is dynamic = 42; sub f() { dd CALLER::<$x> }; f
camelia Int $x = 42
el gatito (** advocate) i can't get lexical vars smh 13:27
lizmat and thus also goes any possibility of optimizing the access to $x out of thwe windos
*window
lexical variables are only visible in the scope they are defined in, and any inner scopes 13:28
el gatito (** advocate) maybe ill inspect the outer call frame and get that variable 13:29
Nemokosch really, what you are trying to do is exactly what dynamic variables are for
el gatito (** advocate) self is not dynamic tho 13:31
Nemokosch that's right, and I'm not talking about that 13:32
el gatito (** advocate) what i want is getting the self variable bruh 13:33
Nemokosch Well I can't relate to that. Seems a bit bizarre, if anything 13:34
el gatito (** advocate) m:perl sub f { dd CallFrame.new(1).my<$x>; } sub g { my $x = 1; f; } g; 13:36
Raku eval Rakudo::Internals::LoweredAwayLexical
el gatito (** advocate) now i only need to get the content of LoweredAwayLexical
m:perl sub f { dd CallFrame.new(1).my<$x>.^methods; } sub g { my $x = 1; f; } g; 13:38
Raku eval (method dynamic (Rakudo::Internals::LoweredAwayLexical: *%_) { #`(Method|3187602760608) ... }, submethod BUILDALL (Any:D: ;; @auto, %init, *%_) { #`(Submethod|3187543758640) ... })
el gatito (** advocate) now im treading into cursed territory lol 13:39
m:perl sub f { dd CallFrame.new(1).my<$x>.dynamic; } sub g { my $x = 1; f; } g;
Raku eval Bool::False
el gatito (** advocate) ok there are no solutions to that i give up 😭 13:41
13:47 QhpAptyj9hj0RQwM joined
lizmat yeah, this will only be solvable with macros 13:48
el gatito (** advocate) or use python 😎 13:56
Nemokosch how do you do this in Python? 🤔
where "explicit is better than implicit" 13:57
el gatito (** advocate) py import inspect def super_(o=None, /): # None = object not specified if o is None: o = inspect.currentframe().f_back.f_locals['self'] # do super logic here 14:02
.f_back gets the outer frame and .f_locals gets the vars in its scope 14:04
Nemokosch yeah this is isn't sane code for sure lol 14:07
lakmatiol well, this is about what python super() does (not quite, since you actually need both the type and the object for super(), but the idea is there) 14:25
the type is in a closure cell, the object is the first argument of the call. 14:26
which is why you can do something like this py def f(__class__): def g(u): a=__class__ return super() return g >>> f(object)(1) <super: object, 1> 14:29
I thought you just did one of callnextsame things in raku though
el gatito (** advocate) that does not guarantee that you will call the method in the superclass it might call another multi candidate in the same class 14:31
lizmat doesn't understand why you wouldn't want to re-dispatch 14:32
Nemokosch super itself kinda violates "explicit is better than implicit" and does so in a weird way
lizmat but I guess I'm suffering from Stockholm syndrome :-)
Nemokosch Remember super().init kinda calls? I doubt it's even possible to argue for the sanity of that 14:34
It hides what it should expose (self) and exposes what it should house (dunder magic)
*hide 14:35
el gatito (** advocate) lizmat: i have demonstrated that several times glot.io/snippets/gidp0msau6 14:38
lizmat of which I said that the observed behaviour might actually be a dispatch bug, but no one has made an issue for it yet ? 14:39
lakmatiol practicality beats purity 14:40
and writing super(__class__, self) in every single piece of code half a dozen times is just kind of silly
Nemokosch It's not practical
el gatito (** advocate) yes it is
Nemokosch Practical would be bare super for the constructor call 14:41
Like in Java
lakmatiol ye, delegating the the super constructor is quite ugly
but alas, no one could invent an adequate pep for that
el gatito (** advocate) i agree
Nemokosch It's the worst of two worlds really
el gatito (** advocate) super.__init__
lakmatiol well, there are other methods that you want to delegate to 14:42
though the ctor is the most common
14:43 rf joined
el gatito (** advocate) super() at this point would be super.__call__ tho (java trap) 14:44
lakmatiol yeah, python isn't a well designed language
but it's limping along somehow
Nemokosch Anyway, this can be done well, as illustrated by e.g Java and Object Pascal
Perhaps C# too, after all, that's a hybrid of these two 14:45
lakmatiol yeah, most languages do just make super a keyword
python... didn't
so you get this weird patch
lizmat m: class A { method a() { dd } }; class B is A { method a() { dd; self.^mro[1].a } }; B.a # how about using .^mro ? 14:46
camelia method a(B: *%_)
method a(A: *%_)
el gatito (** advocate) lizmat: that is exactly what my super does
Nemokosch If somebody threw the mugs yesterday, I'd definitely submit an RFC to forbid methods as a distinct thing 14:47
lizmat m: class A { method a() { dd } }; class B is A { method a() { dd; self.^mro[1].^find_method("a")(self) } }; B.a # more functional
camelia method a(B: *%_)
method a(A: *%_)
Nemokosch Python also rather doesn't have methods, they are just bound subs 14:48
JS doesn't have methods at all 14:49
el gatito (** advocate) you meanperl class A { sub foo(\self) { say 'foo'; } sub bar(\self, $baz) { say "bar$baz"; } }
rf I think Raku has bound subs too in a way
Which imo is the right way to approach methods
el gatito (** advocate) js methods are like python but the self/this is implicit 14:50
Nemokosch But it also has methods that refuse sub semantics often 14:51
Confer slurped named arguments
Also, no straightforward way to reference 14:52
lakmatiol well, a fun difference is that JS binds this at call time, while python binds this at attribute access time.
rf m: class Foo { method bar { 'baz' } }; my Foo $foo .= new; my &m = $foo.^methods[0]; say &m($foo);
camelia baz
rf Thought ^ was pretty neat
el gatito (** advocate) implicit *%_ 😭
lizmat raku.land/github:nxadm/StrictNamedArguments 14:53
el gatito (** advocate) if you don't want to install that package, you can do this method foo(*%()) 14:55
Nemokosch rf: that's metamodel land quite clearly 14:56
el gatito (** advocate) does that make a practical difference
even then method_table is better
rf: ^
Nemokosch Yes, it does make a practical difference 14:57
It really does 14:58
In JS, obj.foo is really just a clueless function
You need to make sure you bind it
el gatito (** advocate) actually yesjs const a = { foo() { return this; }, }; const aFoo = a.foo; console.log(aFoo()); // [object Window] 14:59
js funny
Nemokosch Part of the reason why arrow functions are welcome 15:00
You want to wrap these calls always
So that you don't expose a dangling this
el gatito (** advocate) btw im installing wsl just to publish my Super module 15:01
Nemokosch 🥺 good luck 15:02
WSL is good usually
16:05 guifa_ left 16:14 Heptite joined 16:42 gfldex left 16:52 gfldex joined 16:57 gfldex left 17:00 gfldex joined 17:26 guifa joined 19:12 QhpAptyj9hj0RQwM left 19:31 falsifian joined 19:45 QhpAptyj9hj0RQwM joined 20:15 QhpAptyj9hj0RQwM left 20:23 ab5tract left, ab5tract joined 20:34 ab5tract left 20:35 ab5tract joined 20:45 ab5tract left

Powered by Cro and the Raku® Programming Language.

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

Thank you!