On 2017-12-28 8:38 PM, Emil Velikov wrote: > On 28 December 2017 at 18:05, Drew DeVault <[email protected]> wrote: > Surely you are familiar that weston provides libweston with somewhat > similar functionality, right? > > Since your email is aimed at wayland-devel, it might be better to > point out what part of libweston design does not fit your needs? > > In my (somewhat limited) experience developers seem quite open about > libweston. I cannot see any proposals from you - did I miss them, or > you simply felt uncomfortable posting those to the list?
I briefly addressed our concerns with libweston in the paper. They boil down to problems that generally arise from libweston being refactored out of an existing compositor rather than being made from the ground up as a support library. We didn't find it flexible enough or applicable to niche compositor designs, and we found the codebase rather crufty with lots of design problems from Weston bubbling up through libweston. I had thought the libweston approach is fundamentally flawed and supposed it'd be better to start from scratch. Comparing the two libraries today, I conclude that my assessment was correct. Most who work on wlroots agree that its approach is much better than libweston at all levels. I think libweston is a fine effort for improving Weston in its own right, but for building new compositors the wlroots approach wins out. -- Drew DeVault _______________________________________________ wayland-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/wayland-devel
