Well, I agree this fix would allow non-compliant implementations to keep misusing wl_surface.commit, and that could potentially remove any motivation to find an appropriate resolution for the bug below.
I'm fine with deferring this patch, but I still think applications should not start setting EGL surfaces up before they have been set up by the compositor. Thanks. On Mon, 21 Nov 2016 09:52:40 +0000 Daniel Stone <dan...@fooishbar.org> wrote: > Hi Miguel, > > On 15 November 2016 at 04:49, Miguel A. Vico <mvicom...@nvidia.com> > wrote: > > There is nothing that prohibits the underlying EGL_PLATFORM_WAYLAND > > implementation to attach a buffer or commit surfaces right after the > > Wayland EGLSurface has been created. > > > > Since XDG Shell v6 imposes that no buffer attachments or surface > > commits must be done before a configure is complete, Wayland > > clients shouldn't start setting EGL up until XDG setup is complete. > > > > Related bug: > > > > https://bugs.freedesktop.org/show_bug.cgi?id=98731 > > Per discussion on that bug, I'd like to defer this patch until we > reach a resolution. However, given the pretty firm opinions in this > bug, I'd definitely lean towards retaining the current behaviour, and > doing what we need with EGL implementations (Vivante's does the same > thing, in the opposite direction: it commits too late, rather than too > early!) rather than allow commits to happen outside the user's > control. > > I guess you already know this, but figured I'd better document it for > the list. :) > > Cheers, > Daniel > _______________________________________________ > wayland-devel mailing list > wayland-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/wayland-devel -- Miguel _______________________________________________ wayland-devel mailing list wayland-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/wayland-devel