From: "zverok (Victor Shepelev) via ruby-core" <ruby-core@...>
Date: 2022-12-29T20:28:39+00:00
Subject: [ruby-core:111512] [Ruby master Bug#19278] Constructing subclasses of Data with positional arguments

Issue #19278 has been updated by zverok (Victor Shepelev).


@tenderlove I assure you that I spent a lot of time thinking about possible design decisions, and I believe that while _somewhat unlike_ other classes, the current choice is the best in usability. It is documented (the documentation might emphasize the design choice more, with that I might agree).

The constraints I was working is was:
* flexible creation (e.g. `Point.new(1, 2)` and `Point.new(x: 1, y: 2)` both works)
* as clear error indication with wrong arguments as possible without creating a whole framework for "explanatory errors" (think Rust)
* ease of updating `initialize` protocol for the end user (providing defaults and conversions).

If you have better idea of how to achieve it, I will gladly consider it.

That being said, I believe that "it was surprising on the first pass" does not _necessarily_ equal "ruins usability", and the gains are more than losses.

> Also inheriting from a Data class seems kind of an anti/rare pattern.

I don't think it is 100% true (especially if we'll consider `Data.define(...) do` as an inheritance). At least, even in pre-`Data` times, I have a fair share of usages doing nice things like 
```ruby
class Point < Struct.new(:x, :y)
  def distance_to(other)
    # ...
  end
end
```
I think "immutable value objects" doesn't imply "no additional convenience querying methods".

And, as I said, the `new`/`initialize` pair was **specifically** designed for usability on inheritance (or `do`-block-inheritance). 

When we discussed the initial protocol, Matz was concerned about the ability to provide defaults for _some_ fields (like some additional API maybe, like `Data.define(:x, :y, z: 0)` (which looks nice, but has some grave problems), and I believe that this example provides a clear and usable compromise:
```ruby
class Point < Data.define(:x, :y, :z)
  def initialize(x:, y:, z: 0) = super
end
```
That is intended.

`Data` is a new thing. It has its quirks, but we put a lot into designing it as a small, easy-to-learn, and convenient API.

----------------------------------------
Bug #19278: Constructing subclasses of Data with positional arguments
https://bugs.ruby-lang.org/issues/19278#change-100882

* Author: tenderlovemaking (Aaron Patterson)
* Status: Feedback
* Priority: Normal
* ruby -v: ruby 3.2.0 (2022-12-25 revision a528908271) [arm64-darwin22]
* Backport: 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN
----------------------------------------
I'd expect both of the following subclasses to work, but the subclass that uses positional parameters raises an exception:

```ruby
Foo = Data.define

class Bar < Foo
  def initialize foo:
    p foo
  end
end

class Baz < Foo
  def initialize foo
    p foo
  end
end

Bar.new foo: 1 # Prints 1
Baz.new 1    # Raises ArgumentError
```

I'd expect the subclass that uses positional arguments to work.



-- 
https://bugs.ruby-lang.org/
 ______________________________________________
 ruby-core mailing list -- ruby-core@ml.ruby-lang.org
 To unsubscribe send an email to ruby-core-leave@ml.ruby-lang.org
 ruby-core info -- https://ml.ruby-lang.org/mailman3/postorius/lists/ruby-core.ml.ruby-lang.org/