00:00 kisak: The common answer for why is OpenGL not like modern hardware is usually that it predates those concepts, and evolution at this point is better off in a different API that's closer to how modern hardware works.
00:01 HdkR: Also, GL_NV_command_list
00:58 pac85: Some ancient 3d HW didn't actually have a command processor. Like in a voodoo you draw triangles by poking registers.
00:59 valpackett: hiiii. umm so i'm working on a mipi dsi panel driver and my panel needs to modify the init sequence based on the chosen mode (and also has sequences for live changing refresh rate)..
00:59 valpackett: and it looks like the api currently does not support that at all??
01:01 valpackett: (ignoring the commands and just changing the refresh rate results in artifacts, fwiw)
01:09 Lilla: <valpackett> "hiiii. umm so i'm working on a..." <- is the mode supposed to be set by dt, cmdline or only changed after boot?
01:11 valpackett: changed at runtime, selectable refresh rate in gnome-control-center
01:12 valpackett: (eventually maybe even a userspace pseudo-VRR between fixed modes like android does..)
01:14 valpackett: i can list all the modes, not even the first panel to do that, they appear fine in drm
01:15 valpackett: but in the code that sends the mipi commands to prepare.. there's really no way to see which mode we're preparing for??
02:35 mareko: cssushiman: because nobody thought in 1992 that we would need command buffers
02:39 mareko: GPUs didn't really exist at the time
02:42 mareko: Windows 3.1 had just been released
03:21 mareko: do I understand it correctly that Vulkan has 4 different ways to compile shaders? 1) graphics pipelines, 2) graphics pipelines with dynamic state, 3) graphics pipeline libraries, 4) EXT_shader_object?
08:12 Lilla: Val Packett: perhaps a connector argument could be added to prepare like it was for get_modes https://patchwork.freedesktop.org/patch/344017/
08:13 Lilla: ?
09:46 valpackett: hmm, current mode is reachable via connector?
15:52 Lilla: <valpackett> "hmm, current mode is reachable..." <- crtc, sorry (although that might be actually reachable from connector, but not relevant)
15:53 Lilla: perhaps it could make sense to pass crtc or crtc state here? https://elixir.bootlin.com/linux/v7.0-rc4/source/drivers/gpu/drm/bridge/panel.c#L129
17:11 mareko: zmike: zink flake, the MR has no effect on it: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/95799525