It's a requirement of wlroots, since it uses it to get notified of
Xwayland readiness.

Having o...

This more or less makes everything pretty playable.

We don't want to unhide the cursor in the small period of time where we
don't have a foxued window.

This isn't a perfect fix, see comment.

Output window size, scaling filter, border.

TODO: make output dynamically resizeable, and some ...

That seems to make DC happier when we change the src and dst rects.

Add primitive hold-n-drag.

Ideally we'd get actual MotionNotify and redraw if needed, but probably
grabbed in this case. Th...

Add the concept of a cursor coordinate when that initializes when focusing
a surface.

Put some ...

If a window already has a buffer attached by the time we figure out it
exists and sets its role,...

If the base game is stuck before having rendered anything, we still want
to show overlay layers ...

It's non-fatal, presumably we messed up source coordinates for that flip
but needs root-causing....

Fixes the server going away when we need it most, or when attaching GDB.

So the session script can read it back and know when gamescope is ready.

We're a superset of steamcompmgr now, but have a wider scope, so new
name to reflect it.

If we only take the first commit, we leave a bunch in the queue, which
makes us that much more l...

Not last. Prevents non-vsynced clients from getting too far ahead.

Tries to flip a 1x1 surface, apparently amdgpu doesn't like it

In a liftoff future it would sur...

Now that things are generally more stable, let's take a look.

Use an empty dummy texture for unbound slots, as the driver can't know
what the shader will samp...

This way we never flip anything but our own buffers, to facilitate bug
isolation.

The flip handler could try to deref before we had added the refs.

Also fix the mouse focus edge condition to actually be correct.

Untested for now as it only happens when we lose a race, leaving a
printout to confirm if it wor...

Can't get them to be useful, might need library and amdgpu DC fixes.

Import libliftoff as subproject and plumb our planes through it.

Wait for a flip to complete im...

As opposed to nested.

Thread group size doesn't have to be aligned with image dimensions, as
is the case with 16x16 on...

Only set implicit_sync on in-bound buffers, we'll sync our scanout image
ourselves by exporting ...

We don't want SDL input in the non-nested case at any rate.

So we can easily test single-app scenarios by generating placeholder
ones. Will also make it mor...

That should come in handy for auto-grabbing in the nested case.

It worked well enough with the Big Picture cursor, and seeing it change
scales between Steam and...

Relative mouse broken right now.

We're in sole control of the cursor now, there's no "fake" cursor anymore.

Don't attempt to hid...

As Bas points out, their original WSI allocated them expecting them to
get scanned out, so this ...

Also overload VulkanPipeline_t some more. :/

Change the wl event loop to busy-wait for now as that's easiest to make
unlock periodically.

In order to launch the binary with for example Mesa being installed not into
the regular /usr pr...

The project was licensed under BSD-2-Clause. Add this information back again.

That's a good enough proxy for a real vblank cadence and helps my Envy
not catch on fire.

Some more leading whitespace for good measure.

This should make steamcompmgr messages actually go when we send them.

We're pulling too many fr...

Doesn't help running out right now, as the DMA-BUFs from a new app
frame get imported as differe...

Otherwise we can crash if what we get passed has holes, like when the
main layer loses its textu...

The WL_SURFACE_ID messages force us to register SubstructureRedirectMask,
which means we have to...

Main motivation was to have a clean-ish spot to add a ping-pong
command-buffer instead of alloca...

It causes use-after-free bugs since the current logic is based
on GLX TFP hold/release semantics...

Might as well clean up some of the initial madness.

Caching descriptors is useless when we're importing different images
every draw.

With this and some more fixes, output pipeline looks pretty much perfect.

Working nested Vulkan path now using surface from SDL2 window.

For some definition of working, ...

add some freeing of resources.

frames out of thin air.

Also use dummy ClientMessage events to nudge steamcompmgr's event loop,
...

No texturing yet, that's next after some surgery on the lump of code.

exiting.