-
Notifications
You must be signed in to change notification settings - Fork 138
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
glazier can't compile with both x11 and wayland support #17
Comments
Backtrace of glazier with wayland backend. Crashes on intel and nvidia:
|
Can't build glazier with both backends (I'll file something there too):
|
https://github.com/tauri-apps/tao could also be very relevant here. |
Isn't winit basically a standard by now? In my opinion, using winit is a great idea, why not? |
This has been discussed a few times (on the discord I think). The current reasoning is that glazier is positioning itself to be a better fit for what Lapce and floem will be used for which is desktop applications. Glazier is responsible for handling abstractions over differences in the OS (for things such as timers and menu items). Winit sees that as something that should be implemented as a separate crate and not core functionality. Lapce (and therefore floem) places importance on things like menu items and other similar things working. From the glazier README
This is the answer to the question of why not winit. It does seem like progress on glazier is slow though and I'm doubtful that it will ever pick up traction in the community given the popularity of winit and from my experience support from glazier members is also currently slow. Floem also seems like it would be a good option for mobile but glazier doesn't intend to support it without contributions and without community traction I'm doubtful that will ever happen. Do you still see progress moving forward on glazier @dzhou121? |
This issue with desktop features was actually why I suggested TAO in the first place - it's a fork of winit, but adding all the desktop environment features we need, since Lapce and Tauri have shared needs there. From tao repo:
|
Glazier team is working on a new wayland backend which is promising. TAO wouldn't work for us because GTK backend can't (or not easily) work with Wgpu. |
Thank you for pointing that out - I wasn't aware of that incompatibility issue. It seems like the glazier team already has removed gtk, so it would be interesting to know what the state of the issues initially mentioned by @flukejones are at this point. |
Yes, those are problems, but winit is used by many GUI libraries and is basically the only alive library now. Glazier is getting some work done but should we stick with is because it is more convenient in short term? |
@WeetHet I think sticking with glazier in the short term won't harm Floem. Even if Floem ever does switch away from glazer to winit it isn't going to be much more work then than it will be now (and likely less once problems in the winit scope are solved). Floem currently uses glazier and for the reasons @dzhou121 mentioned switching to winit/tao isn't an option right now without at least a lot of work. Glazier definitely isn't dead. It has the core functionality that is needed including features that are important to Lapce/Floem. The core developers working on Galzier are spread thin working on other problems in xilem and vello. Progress is slow but for now it is a good option for Floem. Going forward it will likely remain a good option for Floem as the goals align well and progress does seem to move well when features are required. |
Glazier seems to have numerous issues on Linux.
a. x11 is unmaintained and no-longer the default graphics session on many distributions
b. it then runs through XWayland, which causes further issues (root cause of Invalid surface on Intel iGPU driver #16 )
I don't know if you want to continue on with glazier - winit is a solid alternative. Or maybe make floem window-management agnostic? Either way the inability to support both x11 and wayland is problematic where wayland should be the default with a fallback to x11.
As it stands the only possible way to run examples or lapce/floem branch on hybrid GPU linux systems is to offload to the dGPU (increasing power use considerably).
The text was updated successfully, but these errors were encountered: