D15512: [startplasmacompositor] Add Wayland socket argument

[Martin Flöser]

graesslin added a comment.


  In D15512#359253 <https://phabricator.kde.org/D15512#359253>, @fvogt wrote:
  
  > > KWin doesn't need this functionality and shouldn't have this functionality and it would be difficult to implement.
  >
  > Why? That goes entirely against what I and probably many others expect from a kwin_wayland call without explicit socket argument.
  
  
  This was a deliberate design decision when writing the code. One reason is that code performing the check and then trying to create the socket is racy. Such a race does not belong into KWin.
  
  Also for me the reason "it was like that in X" is not valid. The only valid question is: what makes sense from a KWin perspective.
  
  > kwin_wayland already checks whether a socket is used, it can't be that hard to extend that to iterate a few socket names.
  
  No kwin_wayland does not check whether a socket is used.
  
  > If you suggest to implement that in a layer higher than kwin, that's not possible as kwin has to create the .lock file itself.
  
  The lock file is not created by KWin, but by the wayland library. That is a layer higher than KWin. If such functionality should exist it should be in libwayland-server and from there available to all compositors. I refuse to accept that every compositor has to implement racy code.
  
  > I'd be fine with this change if it were more generic and allowed to add arbitrary arguments to kwin_wayland.
  
  No other option makes sense to pass to kwin from startplasmacompositor.

REPOSITORY
  R120 Plasma Workspace

REVISION DETAIL
  https://phabricator.kde.org/D15512

To: romangg, #kwin, #plasma
Cc: graesslin, fvogt, plasma-devel, ragreen, Pitel, ZrenBot, lesliezhai, ali-mohamed, jensreuterberg, abetts, sebas, apol, mart
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/plasma-devel/attachments/20181114/f5e814c8/attachment.html>