Introduce Platform.style-name and Platform.os

[tronical]

This replaces the previously hidden StyleMetrics.style-name that was only accessible for internal use, introduced in #8200.

Additionally, this introduces Platform.os.

[tronical]

I tried this manually, but I'm curious about suggestions how to test this. I could "expose" Platform.style-name via property to a test's native code, but then... how to verify that the value is "correct"? I guess I can at least check that it's not empty.

[ogoffart]

Thanks. Looks good to me.

But I don't think we should have Platform without the other stuff like os and stuff. and it's probably not a good idea to rush that in 1.11 so i'd favor waiting.

For the test, maybe you can have something like
out property <bool> test: Platform.style-name != "" && (Platform.os == "macOs" ? platform.style-name == "cupertino" : Platform.os == "windows" ? platform.style-name == "fluent" : true );

or out property <string> style-name <=> Platform.style-name (i wonder if <=> is allowed actually, otherwise just use :) and test that with #[cfg(...)] in the rust test.

For the name, should it be style-name or just style ?

[ogoffart]

I would have left the name as is. But can't hurt.
Maybe a doc comment explaining what is is exactly would be good.

[tronical]

But I don't think we should have Platform without the other stuff like os and stuff.

That sounds good to me. I'll add those "on top" in this PR then.

and it's probably not a good idea to rush that in 1.11 so i'd favor waiting.

Absolutely. I don't intend to propose this for 1.11. As discussed in the API review, I just wanted to do both in one, the change for 1.11 as well as the follow-up, so that it's clearer to see where this goes.

For the test, maybe you can have something like out property <bool> test: Platform.style-name != "" && (Platform.os == "macOs" ? platform.style-name == "cupertino" : Platform.os == "windows" ? platform.style-name == "fluent" : true );

or out property <string> style-name <=> Platform.style-name (i wonder if <=> is allowed actually, otherwise just use :) and test that with #[cfg(...)] in the rust test.

Ah yes, great, with os then it makes sense.

For the name, should it be style-name or just style ?

I prefer style-name. I'd go with that for now and we can review it once more for 1.12 ;-)

[tronical]

Marking this as draft, to make it clearer that I don't intend this for 1.11.

[tronical]

out property test: Platform.style-name != "" && (Platform.os == "macOs" ? platform.style-name == "cupertino" : Platform.os == "windows" ? platform.style-name == "fluent" : true );

Oops, I see that this doesn't work because we go through lengths to make sure that tests are run with the fluent style for consistent results :-). But I'll add this to the widget tests and check at least that the value changes.

[ogoffart]

"iOs" as a uppercase O, should macOs too? and Windows and Linux and Android?

[tronical]

Oops, the values are intended to be lower-case. I'll fix that.

[ogoffart]

Add a note that more values may be added in the future.

I wonder if we should make it an enum.

Also i imagine our user would also want to distinguish between the web and native. We should document that when running on wasm, this does a runtime check.

[tronical]

Add a note that more values may be added in the future.

Ok

Also i imagine our user would also want to distinguish between the web and native. We should document that when running on wasm, this does a runtime check.

I can add that in the docs.

For WASM, we could also add a Platform.is-web-assembly or so (not this particular name maybe, but a boolean - I'm open to suggestions here)