On 25/06/13 12:05, Pekka Paalanen wrote: > Hi, > > yeah, the basic approach seems fine to me. You can get an arbitrary > amount of outputs for testing, if you add this functionality to the x11 > backend. Will try to implement it backend independent if possible.
> It would also be useful to test and guard against crazy(?) > configurations, like multiple position keys in one output section, or > impossibilities like A:left-of-B combined with B:left-of-A or longer > loop. It would be cool to add that to the test suite. The current implementation guards against such kind of loops. The final ordering will be undefined but i think theres not much else you can do about it. > I also see you are trying to cope with dynamic output changes, that's > good, even though Weston on other parts (especially shell) does not > handle it. > > There are some coding style issues, like operator spacing, and line > lengths. > > I didn't review the algorithm, though one thing coming to my mind is, > if you have configured outputs A-B-C in that layout, and only A and C > are present, are they positioned properly? Currently not, as soon as B appears it will be. An interresting problem, will have to think about it. > If you want comments on the code, you should send it inline. We cannot > easily quote an attachment in email. > > Don't worry about the xrandr-replacement for now, it is somewhat > orthogonal to this. > > Primary output is probably more of a shell concept than a Weston core > concept, but I guess you could propose something. I would think that > the core would use it only for the initial position of a pointer, when > a wl_pointer first appears. The shell would do its own stuff if it > wishes. > > > Thanks, > pq > _______________________________________________ > wayland-devel mailing list > [email protected] > http://lists.freedesktop.org/mailman/listinfo/wayland-devel _______________________________________________ wayland-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/wayland-devel
