| zzag | |
| apol |
| KWin |
We hit this if we run an old kwin against new kwayland.
Kwin won't set the name so it's empty, so we'll call
zxdg_output_v1_send_name with an empty string
QtWayland will see that we receive this and update mOutputName to an
empty string which causes client issues.
| Automatic diff as part of commit; lint not applicable. |
| Automatic diff as part of commit; unit tests not applicable. |
The spec says that name and description events will be sent after the xdg-output is created. I believe it might be a problem. We should remove these ifs in KF 5.70.
I interpreted the spec the same way when I wrote this...but we don't have a choice.
I believe it might be a problem
From a QtWayland POV it's fine at least.
We should remove these ifs in KF 5.70.
Why then?
Removing them then would still break Plasma 5.18 and officially we support that combo
Plasma 5.19 will set the name/description, so removing the if won't change anything.
I think we do.
(a) Revert the recent xdg-output patches and land them in KF 5.70 or later. (Plasma 5.19 is going to depend on KF 5.70)
(b) Introduce an API to indicate what version of the protocol is actually implemented by the compositor.
m_xdgOutputManager = m_display->createXdgOutputManager(m_display); m_xdgOutputManager->setVersion(XdgOutputManagerV1Interface::V2); m_xdgOutputManager->create();
I think it's okay to ship the current fix, but only temporarily.
I believe it might be a problem
From a QtWayland POV it's fine at least.
We should remove these ifs in KF 5.70.
Why then?
Removing them then would still break Plasma 5.18 and officially we support that combo
Plasma 5.19 will set the name/description, so removing the if won't change anything.
(a) Revert the recent xdg-output patches and land them in KF 5.70 or later. (Plasma 5.19 is going to depend on KF 5.70)
I don't see how that fixes anything, as Plasma 5.18 should still be support by both KF5.69 and 5.70