Clarify that components' invoke_...() calls through synchronously.
#6457
Conversation
| @@ -83,7 +83,7 @@ pub mod generated_code { | |||
| /// ``` | |||
| pub fn on_hello(&self, f: impl Fn() + 'static) {} | |||
|
|
|||
| /// For each public function declared at the root of the component, a function to call | |||
| /// For each public function declared at the root of the component, a function to synchronously call | |||
There was a problem hiding this comment.
I’m not sure this adds much value. I mean, this is a Rust function declaration, it returns a value, and it’s not declared async. How could it be interpreted as asynchronous?
I’m not opposed to the patch, but I’d like to better understand how this became unclear. Should this perhaps be documented elsewhere, in the general Slint docs? (It applies to all languages)
There was a problem hiding this comment.
It's functions to interface with the Slint context, which is kind of a dark pit without deeper knowledge. Okay, invoke_...() for function returns a value; but not invoke_...() for callback [in the sample component]. You could argue that something like changed-callbacks are also deferred. I just found myself in the situation of not trusting how exactly it behaves, when needing exact information to be able to prevent RefCell borrowing panics. And the docs didn't provide a guarantee in this regard. (See also #5885.)
There was a problem hiding this comment.
But even regarding functions: I had function that takes a string and returns nothing. Because of that, I went directly to the docs text, ignoring the invoke_...() signature of the sample component, because I deemed it irrelevant and didn't think of other possible signatures.
There was a problem hiding this comment.
it returns a value, and it’s not declared async. How could it be interpreted as asynchronous?
It doesn't always return a value and people wouldn't interpret it as async (rust term) but maybe as queued.
I think this clarification is accurate so I'll merge this patch. Thanks
This makes it clear that you can expect that the Slint code isn't called in the next event loop run or so, but immediately.
Is this factual?