ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver || https://docs.mesa3d.org/drivers/zink.html
<fdobridge> <g​fxstrand> Okay. Terrible hack works. Now I have to figure out how to do it properly.
<fdobridge> <g​fxstrand> I'll figure that out tomorrow.
<fdobridge> <z​mike.> If you post your hack I can probably tell you the proper fix to save you time
<fdobridge> <g​fxstrand> ```
<fdobridge> <g​fxstrand> diff --git a/src/egl/drivers/dri2/platform_x11.c b/src/egl/drivers/dri2/platform_x11.c
<fdobridge> <g​fxstrand> index fd669e1cc67..c524dfd8c7e 100644
<fdobridge> <g​fxstrand> --- a/src/egl/drivers/dri2/platform_x11.c
<fdobridge> <g​fxstrand> +++ b/src/egl/drivers/dri2/platform_x11.c
<fdobridge> <g​fxstrand> @@ -1094,6 +1094,19 @@ dri2_x11_swap_buffers_msc(_EGLDisplay *disp, _EGLSurface *draw, int64_t msc,
<fdobridge> <g​fxstrand> }
<fdobridge> <g​fxstrand> #endif
<fdobridge> <g​fxstrand>
<fdobridge> <g​fxstrand> +#include "dri_drawable.h"
<fdobridge> <g​fxstrand> +
<fdobridge> <g​fxstrand> +static void
<fdobridge> <g​fxstrand> +kopper_x11_check_surface(struct dri2_egl_surface *dri2_surf)
<fdobridge> <g​fxstrand> +{
<fdobridge> <g​fxstrand> + struct dri_drawable *draw = dri2_surf->dri_drawable;
<fdobridge> <g​fxstrand> + if (dri2_surf->base.Width != draw->w || dri2_surf->base.Height != draw->h) {
<fdobridge> <g​fxstrand> + dri2_surf->base.Width = draw->w;
<fdobridge> <g​fxstrand> + dri2_surf->base.Height = draw->h;
<fdobridge> <g​fxstrand> + dri_invalidate_drawable(draw);
<fdobridge> <g​fxstrand> + }
<fdobridge> <g​fxstrand> +}
<fdobridge> <g​fxstrand> +
<fdobridge> <g​fxstrand> static EGLBoolean
<fdobridge> <g​fxstrand> dri2_x11_swap_buffers(_EGLDisplay *disp, _EGLSurface *draw)
<fdobridge> <g​fxstrand> {
<fdobridge> <g​fxstrand> @@ -1108,6 +1121,7 @@ dri2_x11_swap_buffers(_EGLDisplay *disp, _EGLSurface *draw)
<fdobridge> <g​fxstrand> */
<fdobridge> <g​fxstrand> kopperSwapBuffers(dri2_surf->dri_drawable,
<fdobridge> <g​fxstrand> __DRI2_FLUSH_INVALIDATE_ANCILLARY);
<fdobridge> <g​fxstrand> + kopper_x11_check_surface(dri2_surf);
<fdobridge> <g​fxstrand> return EGL_TRUE;
<fdobridge> <z​mike.> Oh god
<fdobridge> <z​mike.> I don't do egl after noon so that's gonna have to wait
<fdobridge> <g​fxstrand> You asked.
<fdobridge> <z​mike.> Friends Don't Let Friends Do EGL.
<fdobridge> <g​fxstrand> Friend don't let friends debug X11 and yet no one stopped me! 😝
<fdobridge> <z​mike.> I can only hope you're drinking heavily now
<fdobridge> <g​fxstrand> Merry Christmas: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34015
<fdobridge> <g​fxstrand> @tiredchiku ^^
<fdobridge> <O​wo> Single-pixel buffers don't exist in x11, right?
<fdobridge> <O​wo> Like, the Wayland protocol equivalent
<fdobridge> <g​fxstrand> Sorry to wake you. Go back to sleep. You can test whenever you get around to it.
<fdobridge> <S​id> oh lol no
<fdobridge> <g​fxstrand> Not sure what you mean. You can have a 1x1 window.
<fdobridge> <S​id> you didn't wake me 😅
<fdobridge> <g​fxstrand> Hehe. Good.
<fdobridge> <O​wo> Wrote it while occupied, might've not been fully present. I mean asking the server to give you a single-color buffer of one pixel, so you don't have to fill it yourself.
<fdobridge> <O​wo> Probably not related
<fdobridge> <O​wo> Reading the protocol here, it would be resized after creation to the desired size, and then drawn on, so it's never 1x1 during presentation.
<fdobridge> <g​fxstrand> Ah. I'm not sure. I mean, you can fill a region with the X render protocol. Not really the same thing but accomplishes something similar? In any case, that's all unrelated to the window creation race we were hitting with Firefox.
<fdobridge> <O​wo> Yeah
<fdobridge> <S​id> this works good :kittynod:
<fdobridge> <S​id> time to bisect xserver for the cursor thing
<MoeIcenowy> redsheep: technically I am the current investigator of new Zink + IMG closed Vulkan
<MoeIcenowy> IMG closed Vulkan is just some piece of crap
<MoeIcenowy> you will easily get trapped by asserts and crash
<MoeIcenowy> especially for untraditional shaders (e.g. ones translated by Zink
eukara has joined #zink
funderscore is now known as f_
<fdobridge> <g​fxstrand> Remind me why I just spent 2 weeks fixing KWin X11?
<fdobridge> <m​ohamexiety> oof
<fdobridge> <r​edsheep> I'd assume given how long plasma 5 was around that plasma 7 ia quite a ways off
<fdobridge> <r​edsheep> I'd assume given how long plasma 5 was around that plasma 7 is quite a ways off (edited)
konstantin has joined #zink
<fdobridge> <S​id> yeah, we got up to plasma 5.25, afaik
<fdobridge> <S​id> plasma 7 will probably be on qt7
<fdobridge> <e​rmine1716> There are no (public) plans for qt7
<fdobridge> <S​id> yet
Alex___ has joined #zink
Alex____ has joined #zink
Alex____ has quit []
Alex___ has quit [Remote host closed the connection]
eukara has quit []
eukara has joined #zink