Dealing with copyWith nesting

[MoathCodes]

Hello,
I love this package so much I use it for almost all of my project, however one point that felt a bit annoying is when customizing something small like a padding or a color for just a specific widget I have to go into a deep nest of copyWith statements, a while ago i saw a post on reddit of a package called draft, and i was wondering if something like this could be implemented into forui.

in the reddit post the author gave this example:

@draft
class Foo { ... }

final foo1 = Foo(...);

// modify it using draft
final foo2 = foo1.produce((draft) {
 draft.list.add(1);
 draft.b.c = 1;
})

// the old way using copyWith
final foo2 = foo1.copyWith(
 list: [...a.list].add(1),
 b: a.b.copyWith(
  c: a.b.c.copyWith(
   value: 1,
  ),
 ),
)

and this is one example in my code:

         style: (style) => style.copyWith(
              decoration: style.decoration.copyWith(color: colors.barrier),
              unselectedLabelTextStyle: style.unselectedLabelTextStyle.copyWith(
                color: colors.secondaryForeground.withAlpha(150),
              ),
            ),

I think implementing something like draft would make editing styles much more pleasant.

[Pante]

We investigated a similar approach based on Draft awhile ago! I think that while this approach has some nice properties such as handling nullable types better, it only produces marginally better code:

• It doesn't scale well if you're editing multiple deeply nested properties at once, if I remember correctly, you'll actually end up with slightly more verbose code because you can't use the cascading syntax (foo..bar) and have to either assign it to a local variable or repeat it.

• Draft doesn't mesh with types defined in Flutter, e.g. TextStyle, you'll still have to perform a copyWith(...). You'll end up with a mix of Draft + copyWith. We could technically create an extension on those types but I'm not too sure how I feel about that.

All that said, I agree that copyWith is far from ideal and pretty verbose and we should probably do something about it. I'm just not certain what that "something" is just yet. We're also actively looking into improving the theming system as a whole (again), with better support for variants and states.