zzag added a comment.
In D28882#658780 <https://phabricator.kde.org/D28882#658780>, @apol wrote: > > In future, it might be faster to put up just the interface xml for review first. > > @davidedmundson @zzag I don't really see how it would have made a difference, you only decided to review it 7 to 10 days after I first submitted it. 🤷 > > > What about using existing wl_output objects? The add_source event can be simplified quite a lot (even maybe dropped). For windows, we could use string handles. > > So your proposal would be to pass a random resource id or string and see what the compositor spits back? Well, it's not that different from passsing uints like this patch proposes. In either case we have to have some protection against garbage input values and possibly inert resources. wl_outputs are not some random objects and practically every GUI application keeps track of them. > My intention was to keep the protocol auto-contained to some extent so every client doesn't need to track windows and screens to be able to request a stream. REPOSITORY R127 KWayland REVISION DETAIL https://phabricator.kde.org/D28882 To: apol, #kwin, jgrulich, davidedmundson, zzag Cc: meven, davidedmundson, romangg, zzag, kde-frameworks-devel, LeGast00n, cblack, michaelh, ngraham, bruns
