Here's some notes from the breakout sessions we had yesterday for my own recolation.
KWin and Wayland on Kubuntu
- Differences between Wayland and Mir are minimal, only client vs server side buffer allocation, so no benefit from it for a desktop distro
- dri3000 is a project to make X do client side buffer allocating
- KWin won't support Mir, (unless canonical wants to do it) it has no ABI stability
- KWin was working on Wayland suppport, now shifted effort to Qt 5 support before going back to Wayland work.
- KWin in 4.11 might work with Qt 5 but not for distros to use
- X wayland is a rootless X server used in both Wayland and Mir, no use for KWin which needs a rooted X server.
- wayland has libwaylandserver and libwaylandclient libraries, libwaylandclient for use in toolkits like Qt.
- current KWin architecture not fully planned but probably Weston as system compositor and KWin as client compositor
- worry that Canonical might havily patch mesa for Unity/Mir making Kwin/Weston break (more then it already is)
- Only Weston needed for LightDM then Kwin started same as currently by startkde (maybe a bit earlier in the script)
- Change to Wayland for Kubuntu about 14.10 or 15.04
- Wayland & Weston release every 3 months, KWin every 6 months
- Unity will have 1 binary equivalent to X, compiz and unity
- Weston can be run on X for testing
- current thinking is to ensure Debian's Weston packages are in Ubuntu archive and use those for Kubuntu, might make ubuntu-desktop and kubuntu-desktop not easily co-installable
- Installer ubiquity-dm a tricky point
- KDE people get frustrated when their point releases don't get into distros or get in very slowly
- ubuntu SRU process is beurocratic for a reason - introducing new bugs is really nasty
- lightdm did introduce a new bug recently in a point release which was caught in the SRU process
- building SRUs in a PPA first to test unnecessary and time wasting, just test it locally then upload to -proposed
- possible scope to have approved from ~ubuntu-sru during the 7 day testing period rather than before to remove one blocker?
- delay in KWin updates causes bug reports to keep coming into KWin bugzilla, very time consuming for upstream
- work items too brief - danger of not understanding later what a point is
- use of trello to be encouraged for work item tracking with enough info to understand each point from someone coming fresh
- UEHS unknown if it works reliably, needs to be investigated and improved to update packaging as much as possible itself, at lp:dehs
- public kde-packager e-mail list (discussion has been done several times in past)
- automated copyright file from upstream?