Use raw-window-handle version 0.6 #132
Conversation
There was a problem hiding this comment.
Implemented Web, built on top of forkgull/winit#1 now, until that's merged.
|
We should probably add feature flags for raw-window-handle versions, like in rust-windowing/winit#3126. Edit: Though I guess it's a bit harder to do with adaptive features for arguments. Without some trait magic. If we can support both 0.5 and 0.6 in a clean way, that would be good though. |
|
Actually, I guess we can't have a function that accepts either something implementing both Anyway, if it's going to support only 0.6, it should be released alongside the Winit version that adds support for 0.6. |
Signed-off-by: John Nunley <dev@notgull.net> Co-Authored-By: dAxpeDDa <daxpedda@gmail.com>
Signed-off-by: John Nunley <dev@notgull.net>
notgull
force-pushed
the
notgull/rwh-v0.6
branch
from
fc19d27 to
1f44e3a
Compare
Signed-off-by: John Nunley <dev@notgull.net>
Signed-off-by: John Nunley <dev@notgull.net>
notgull
requested review from
ids1024,
jackpot51,
madsmtm and
john01dav
as code owners
|
CI seems to be failing because the benchmark fails to build. So that also needs to be updated. |
Signed-off-by: John Nunley <dev@notgull.net>
| pub struct WaylandDisplayImpl { | ||
| conn: Connection, | ||
| pub struct WaylandDisplayImpl<D: ?Sized> { | ||
| conn: Option<Connection>, |
There was a problem hiding this comment.
Instead of using an Option, I'd probably just rely on field order, and have a safety comment indicating this requirement.
There's also ManuallyDrop, though the docs for it seem to just recommend using field order for this sort of thing instead of dealing with it: https://doc.rust-lang.org/std/mem/struct.ManuallyDrop.html#manuallydrop-and-drop-order
There was a problem hiding this comment.
I'd rather rely on Option than unsafe ManuallyDrop or fragile field ordering.
| pub unsafe fn new(display_handle: WaylandDisplayHandle) -> Result<Self, SoftBufferError> { | ||
| // SAFETY: Ensured by user | ||
| let backend = unsafe { Backend::from_foreign_display(display_handle.display as *mut _) }; | ||
| impl<D: HasDisplayHandle + ?Sized> WaylandDisplayImpl<D> { |
There was a problem hiding this comment.
Why ?Sized? It wouldn't actually be possible to call new with a display: D is D isn't Sized, right?
There was a problem hiding this comment.
This way it allows for Context<D> to be coerced into Context<dyn HasDisplayHandle>, and be used like that.
| @@ -1,3 +1,5 @@ | |||
| // TODO: Once winit is updated again, restore this test. | |||
There was a problem hiding this comment.
Winit is released, so presumably this can be updated. Though https://github.com/notgull/winit-test will need a release.
…Close` again It seems rust-windowing#132 included a copy-paste typo to trigger draw logic on `CloseRequested` instead of on `RedrawRequested`. Signed-off-by: Marijn Suijten <marijns95@gmail.com>
…Close` again It seems rust-windowing#132 contains a copy-paste typo to trigger draw logic on `CloseRequested` instead of on `RedrawRequested`. Signed-off-by: Marijn Suijten <marijns95@gmail.com>
…Close` again It seems #132 contains a copy-paste typo to trigger draw logic on `CloseRequested` instead of on `RedrawRequested`. Signed-off-by: Marijn Suijten <marijns95@gmail.com>
cc rust-windowing/raw-window-handle#125, rust-windowing/winit#2943
Testing out the new release before it becomes official.