This channel is intended for people just starting with the Raku Programming Language ( Logs are available at
Set by lizmat on 8 June 2022.
00:16 jgaz left 00:28 deoac left 02:20 Manifest0 left 02:28 Ebudae joined 02:29 Heptite left 03:08 Ebudae is now known as Heptite 04:04 kjp left 04:05 saint- left 04:07 kjp joined 04:50 rf left 05:32 Heptite left 08:39 QhpAptyj9hj0RQwM joined 10:28 Manifest0 joined 11:14 frost joined
snonux Hi there, how can I call the same method of a parent class? E.g. in: class Foo { method test { <Foo.test>.say; } }; class Bar is Foo { method test { ALSO-CALL-test-FROM-FOO-HERE(); <Bar.test>.say; } };; 11:21
Was looking at 11:22
lizmat is what you probably want to see 11:23
Nemokosch hi, from what I learned, this is a more complex topic than it sounds at first - there are several methods a call could fall back to, including the possibly multiple ancestor classes and mixins. If you are confident what your parent class is, a simple way would be to do something like self.Foo::test 11:24
snonux @lizmat, ahh, yes! callsame() works, thanks!
Nemokosch callsame is supposedly the clean solution but it's really for much more than a method chain - it's a mixture of method resolution and multiple-dispatch resolution (based on parameters) 11:25
lizmat if it works, it works, no ? 11:26
snonux thanks Nemokosch, yes, in my case, i only have one parent class and no roles mixed in. 11:27
Nice, that it's actually 'callsame' to be called - which is pretty close to the question I asked 'how can I CALL the SAME method'. What you write is what you get :-) 11:31
lizmat snonux: indeed, that was one of TimToady's design goals 11:32
Nemokosch yes, it works - I'm just saying, don't be surprised that you get a bit more than what you think you are asking for 11:33
there can be times when that's exactly what you desire and times when it won't (easily) do the trick 11:34
el gatito (** advocate) self.^mro[1].^method_table<test>(self) will always call the superclass method 😎 12:19
Nemokosch 😂 12:27
lizmat Note that this may depend on Implementation details and may stop working at any time 12:31
and as such not really beginner's stuff 12:33
12:39 QhpAptyj9hj0RQwM left
el gatito (** advocate) the metamodel is implementation detail? 12:54
12:59 QhpAptyj9hj0RQwM joined
lizmat well, it is a grey area: there are no spectests for method_table, and no Rakudo tests for it either 12:59
even though it is documented... so there's that
and we recently had some nasty module regressions when some aspect of the metamodel needed changing (an extra parameter to a method) 13:03
el gatito (** advocate) there are cases where redispatching does not give expected behavior like this 13:15
in those cases you have no choice but this 13:16
actually you have another choice of self.B.f 13:18
but still
lizmat you mean self.B::f ? 13:23
but yes, *that* would be preferable way of doing and *that* is sure to remain supported
el gatito (** advocate) then if i want to call it, do i have to include self in the argument? 13:29
like self.B::f(self: #(...))` 13:30
lizmat not in that case, as it is still considered to be a method call 13:33
m: class A { method b($a) { dd $a } }; A.A::b(42) 13:34
camelia 42
el gatito (** advocate) it is sad when raku’s 3 ways of calling superclass method is either: - not really super (redispatch) - implementation dependent magic - just spelling out the class name (explicit is better than implicit i guess?) 13:43
Nemokosch thank heavens that I didn't have to say that... 13:44
you might have social credits left to be taken with good faith
lizmat why is redispatch not really super ? 13:48
el gatito (** advocate) this 13:49
lizmat m: class A { method a() { dd } }; class B is A { method a() { dd; nextsame } }; B.a # this ? 13:51
camelia method a(B: *%_)
method a(A: *%_)
lizmat m: class A { method a() { dd } }; class B is A { method a() { callsame; dd } }; B.a # this ?
camelia method a(A: *%_)
method a(B: *%_)
el gatito (** advocate) have you read the code 13:52
lizmat is looking 13:59
Nemokosch the other thing is that I kept asking about and never got a clear answer is: is there a standard, non-metamodel way to get hold of a method, in the first place? 14:05
or only find_method, lookup sort of stuff?
lizmat so, a part of the complexity in that example is that A_ is a subclass of A
given that, I think the dispatch is correct 14:06
now, there may be reasons for having an A_ inheriting from A, but it feels to me that using roles may be a better solution 14:07
now, the interesting part is when A_ is *not* a subclass of A
el gatito (** advocate) > given that, I think the dispatch is correct yes, but it is not the behavior given by the mro version
the mro behavior is the expected one
lizmat in that case, the calls are: C: hi A_, B: hi A_, B: hi A_
el gatito (** advocate) yes 14:08
lizmat which I think is actually an error, having B: hi A_ in there twice
so if you could make an issue for that, please
el gatito (** advocate) but the redispatch is C: hi A_ B: hi A_ C: hi A B: hi A
lizmat as to Nemokosch's question 14:09
snonux i see, you people are still ongoing with the dispatch :D
el gatito (** advocate) isn't that D: hi A_ B: hi A_ B: hi A
lizmat snonux: we know there are some long standing issues with dispatch, but fixing them is *hard* and may affect performance drastically 14:10
as just about everyting in Raku is dispatch
el gatito the only thing I changed in the code, was remove the "is A" 14:11
el gatito (** advocate) about that idk why lol
also dispatch isn't the problem in this case the problem is raku doesn't have a convenient way to truly delegate method calls to the superclass no super or anything even tho that should be pretty easy to implement 14:16
lizmat well, you're welcome to create a PR or a module :-) 14:17
14:51 ab5tract left 15:20 frost left 16:04 Heptite joined 16:20 yjh left 16:42 QhpAptyj9hj0RQwM left 19:05 QhpAptyj9hj0RQwM joined 19:43 QhpAptyj9hj0RQwM left, QhpAptyj9hj0RQwM joined 20:45 QhpAptyj9hj0RQwM left
Yeuph Would someone be able to give me some pointers? I'm trying to figure out how to get a class to accept sets What would that look like? I've been tinkering for a while and documentation is limited 20:58
like `class Sets { has [what goes here?] } And how to add that to the class 20:59
Nemokosch has is for creating attributes and properties - so far so good? 21:01
Yeuph assuming that SetHash is included in attributes and properties, yes
(also yes if not; which would mean that is the route of my issue) 21:02
Nemokosch I mean, you know what you are going for here? To put a SetHash inside a class so that the class can contain it for whatever purpose 21:03
well, the object instances can
Yeuph I do know what I want to make, I just haven't used Raku in about a year at this point and can't claim to know the language well. I'm coming from C++ 21:04
which probably has near-zero translation value
Nemokosch so this is like private/public: field declaration
Yeuph ok
Nemokosch is this what you want to achieve?
Yeuph public, yes
Nemokosch okay, so you can do has SetHash $.foobar 21:05
you don't need to declare the type, sometimes it's actually better to avoid it
but this is the syntax
Yeuph alright, then declaring that set in the class is done how? `class Set { has SetHash $.foobar } my &thing = => [how are we approaching this?]) 21:08
Nemokosch this will create a private attribute $!foobar and an accessor method foobar to it
first off, I wouldn't overload the name Set if I were you, it might bite back somewhere 21:09
Yeuph yeah my names are separate in what I'm actually writing
(which isn't much at this point and the code isn't worth posting)
Nemokosch second, storing it into a &variable (so a Callable) probably won't work, unless you make your class implement Callable
and I doubt that's your intention anyway 21:10
Yeuph I should probably breeze through some of the Raku course stuff for a few hours
I'm gonna do that. I was hoping I'd remember stuff/be able to just figure it out. I need to just spend a few hours going over the language 21:11
what's the best documentation/course now? 21:12
Nemokosch and third, actually, I'm not sure what you can feed in there as the value of the named argument; my guess would be, anything that can be on the rhs of an assignment of a variable like my SetHash $foobar
so typically something coerced into SetHash
Yeuph ok 21:13
is there a website for Comma error messages? 21:14
Nemokosch I'm afraid you'd need to ask people with Comma affiliations for that 21:15
Yeuph ok thanks for your time
Nemokosch for classes, I'd still say this
shmup also Yeuph idk if you've used the search feature on the docs, but it's a pretty nice breakdown of things and i've been using it a lot: 21:18
Nemokosch the search can usually make an "educated guess" where a certain topic is covered 😛 21:19
shmup well i actually would have searched for this which is even better results :P 21:20
ive also been using chatgpt as a feedback loop on asking questions about syntax i see in code, and it generally helps me with looking the next thing i wanna look up in some doc/etc 21:21
it really does have parallels to a question you might drop in #irc if the outcome (often) is giving you a thing to specific thing/direction to research 21:23
i am not able to type sentences rn :) 21:24
Yeuph alright I'll play with this as well 21:25
22:02 rf joined 22:16 QhpAptyj9hj0RQwM joined 22:56 QhpAptyj9hj0RQwM left