fix(windows): set physical values in bounds method

[kanatapple]

wry/src/webview2/mod.rs

Lines 1193 to 1198 in f56ae29

	bounds.position = LogicalPosition::new(position_point[0].x, position_point[0].y).into();
	bounds.size = LogicalSize::new(
	(rect.right - rect.left) as u32,
	(rect.bottom - rect.top) as u32,
	)
	.into();

bounds.position/size expects Logical position/size, but position_point and rect are Physical values.
So conversion is necessary.

closes tauri-apps/tauri#10053

[wusyong]

Thanks for helping fix the bug! Could you a change file before I can merge it?

[kanatapple]

@wusyong
I added a change file.

[amrbashir]

we don't need these, since we can just use Physical position without converting to logical

[amrbashir]

no need to convert to logical

[kanatapple]

@amrbashir
tauri-apps/tauri#10053

Have you seen this issue?
This is caused by not converting to logical.

[amrbashir]

the field position and size in Rect are of type Position and Size which can be constructed directly from Physical or Logical type, it is the responsibility of the user to convert the position/size later to their preferred variant.

[amrbashir]

you can do

Rect {
	position: PhysicalPositon::new().into(),
}

as well as:

Rect {
	position: LogicalPosition::new().into(),
}

and the user of .bounds() function can convert to their needed types:

let bounds = webview.bounds()?;
let position = bounds.position.to_logical(scale_factor);

[kanatapple]

Is this what you mean?

bounds.position = PhysicalPosition::new(position_point[0].x, position_point[0].y).into();

The original code is wrong because it uses logical, right?

bounds.position = LogicalPosition::new(position_point[0].x, position_point[0].y).into();

[amrbashir]

yeah this is what I mean

[amrbashir]

I still haven't tested this so I am not sure yet if the correct value should be LogicalPosition or PhysicalPosition or maybe the bug could in set_bounds not bounds where it should be passing logical values to win32 APIs instead of physical.

[pewsheen]

I got a bit confused by the definition of Logical lol

- (set) SetWindowPos in set_bound should be Logical, meaning it should have DPI awareness.

• (set) SetWindowPos in set_bound should be Physical, since our app is DPI unawareness.
• (get) The position we get from GetClientRect should be Physical since our app is DPI unawareness. (not quite sure)

- So the LogicalPosition for SetWindowPos should be the PhysicalPosition from MapWindowPoints * dpi_scale_factor, which gets a larger value if the DPI is higher.

It looks different from macOS's implementation because macOS will get a smaller coordinate if the DPI is higher.

[pewsheen]

I'm not sure what we get from WIN32 API is already scaled now 😢 , but I think we should use Logical in set_bounds

Update: oh, I forgot we are DPI unawareness, so it should be Physical, Windows will scale it by the DPI factor

[amrbashir]

no need to convert to logical

[kanatapple]

@amrbashir
I just set physical values without converting.

[kanatapple]

Sorry I don't understand the process, is there anything else I should do?

[kanatapple]

@amrbashir
Is there anything else I should do? (except fixing covector)
Sorry to bother you, but please check.

[amrbashir]

Apologies, I got busy with some other stuff, I will review and test today.

[amrbashir]

Thank you

[github-actions]

Package Changes Through 566b53b

There are 1 changes which include wry with patch

Planned Package Versions

The following package releases are the planned based on the context of changes in this pull request.

package	current	next
wry	0.41.0	0.41.1

Add another change file through the GitHub UI by following this link.

---

Read about change files or the docs at github.com/jbolda/covector

[kanatapple]

@amrbashir
Thank you for the review and merge!