All committers have signed the CLA.

Thank you for your contribution and the pull request. I really appreciate your effort.

I must say, your idea of serializing using __enum<> is quite clever. However, I would prefer a less hackish approach, ideally just saving the string representation of the enum.

To achieve this, I believe there are two possibilities that we can explore:

1. The cleanest approach would be to determine the type of the property so that we can generate an EnumValue from a string. We can identify the type by utilizing the properties_and_callbacks() function on component. It would be helpful to cache it in a HashMap and then pass the expected type to the from_json function. This approach is somewhat similar to what we do in the JS integration:

   slint/api/node/native/lib.rs

   Line 155 in c8185a9

   fn to_eval_value<'cx>(

   The advantage of this approach is that it would allow us to serialize other types that can be deserialized from a string, such as images (from their filename), colors, etc. However, I understand that this may be more difficult.

2. An alternative approach would be to modify the code to accept a Value::String(...) wherever we currently expect a Value::EnumerationValue(...). In other words, we would allow assigning a Value("bar") for an enum that has "bar" as one of its values.

From looking at the code, it appears that these modifications would not be required in many places:

• In the TyFrom<enum> implementation:

  slint/internal/interpreter/api.rs

  Line 278 in c8185a9

  impl TryFrom<Value> for i_slint_core::items::$Name {

• In the comparison logic:

  slint/internal/interpreter/api.rs

  Line 137 in c8185a9

  impl PartialEq for Value {

• In the check_value_type function that verifies the correctness of the value:

  slint/internal/interpreter/eval.rs

  Line 1200 in c8185a9

  Type::Enumeration(en) => {

(As a side node, we might want to change Value::EnumerationValue from its current form. Instead, it would be more efficient to represent it with an integer as indicated by the "FIXME" comment where it is declared, but that's out of the scope of this PR)

Once again, I want to express my gratitude for your pull request. I hope my suggestions and explanations make sense to you. Please let me know if you have any further questions or if there's anything else I can assist you with.

I went for the first solution.

I'm not entirely sure about the error handling, but it works.

By the way, thank you for the detailed briefing on how to do it well !