Refactor impl Default generation
#1446
Conversation
|
I know a gir rework is ongoing, so I am not sure if refactoring PRs makes sense. The reason for this is to tackle gtk-rs/gtk-rs-core#975 by adding a new |
|
gtk3 fails because it has which is now generated... on the one hand I guess it is a good thing we can drop manual code, on the other it is not clear to me why this was not happening before :) |
| let name = &analysis.name; | ||
| if let Some(func) = analysis.default_constructor() { | ||
| general::declare_default(w, env, name, func) | ||
| } else if has_builder_properties(&analysis.builder_properties) { |
There was a problem hiding this comment.
The reason why with this patch RadioButton etc get picked up is because before we were falling back to object::new only if there was a new() function, but it did not have 0 params.
That seems rather odd and with the new code I did not get any false positive in core and gtk3
Move the detection of a default constructor function in the analysis. Move the logic about using Object::new into object.
| // For now we only look for new() with no params | ||
| && f.name == "new" | ||
| && f.parameters.rust_parameters.is_empty() | ||
| // Cannot generate Default implementation for Option<> |
There was a problem hiding this comment.
Should also do the same for Result?
|
Would be nice to land this. But it needs PRs in at least 2 out of gtk-rs-core/gstreamer-rs/gtk4-rs with the resulting changes. |
Move the detection of a default constructor function in the analysis. Move the logic about using Object::new into object.