[RFC wayland] protocol: Add wl_surface.buffer_damage

[Pekka Paalanen]

On Mon, 9 Nov 2015 10:28:13 -0800
Bill Spitzak <spitzak at gmail.com> wrote:

> I believe wl_surface.damage should be defined as "the same as
> w_surface.buffer_damage if the only transforms are due to an integer buffer
> scale and the 8 possible buffer transforms". This means it is undefined if
> the wl_scalar proposal is used, and avoids the need to define it for any
> future transform possibilities.

I have no idea how that could help with anything. It does not matter
what the mapping is between surface and buffer coordinate systems:
damage is in surface space, and buffer_damage is in buffer space. There
is nothing ambiguous about that, even if someone added even more
transformations between them.

> For very similar reasons I believe the wl_scalar and the subsurface apis
> will need to be changed to be in buffer coordinates. So a defined suffix
> (such as _buffer or _pixels) is probably a good idea so these can be fixed
> in a similar way.

That leads to fractional "pixel" units inside a compositor, wl_viewport
would then completely bypass buffer scale and transform (transform is
not representable in wl_viewport), and I simply don't see the point at
all.

Please provide a real-world use case that would benefit from what you
propose. Nothing that you "can imagine", but a real case that exists,
just like eglSwapBuffersWithDamage is a real use case for buffer_damage.

For example, showing that GStreamer will have problems using
wl_viewport properly on a hidpi output would be valuable right now when
the scaler extension is not yet stable. Mind, that you cannot show this
unless you know how Gstreamer works. Speculation does not count.

This is also off-topic for this thread, so please start a new one about
the scaler extension.


Thanks,
pq
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 811 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/wayland-devel/attachments/20151110/4a18ae39/attachment.sig>