esjeon added a comment.

  In D26338#588960 <https://phabricator.kde.org/D26338#588960>, @davidedmundson 
wrote:
  
  > On the conceptual level, does it make more sense for the relevant client 
(konsole) to monitor for compositing changed itself. It's something that could 
change at runtime regardless, so it needs to react to it anyway?
  >  If it does need that code anyway, there's no point doing fixes in startkde
  
  
  Technically, yes, apps can do that. But this will force changes to any apps 
that embed Konsole (i.e. dolphin, kate) or depend on desktop effects. Also, 
this issue will arise whenever an app implements such features. We can just 
solve this here and forget about it forever.
  
  Also, the problem only occurs during session restoration, so it's clear that 
one change here can fix for all applications.
  
  > Code wise, this gives a timeout for anyone who doesn't run kwin composited, 
which would be a problem as is.
  
  Welp, I agree, but there's just no way to figure out if user wants compositor 
or just WM. A possible solution is to make this toggleable through 
configuration or env, which isn't exactly beautiful.
  
  Still, the impact can be minimized by moving `wmjob` to right before session 
restoration instead of stage0.
  
  > also I don't like the split that ksmserver is waiting for kwin to register 
on ICE, and startkde is waiting on kwin through a combo of both DBus and X 
atoms in KWindowSystem::compositingActive(). That's 3 different paths all used 
at once.
  > 
  > I've been wanting to move the kwin startup from ksmserver into startkde 
anyway, doing those things together would make the most sense.
  
  I agree. Things will be much cleaner if WM is handled in startkde. WM isn't 
really a part of session, and ksmserver already treats WM as a special case 
anyway.

REPOSITORY
  R120 Plasma Workspace

REVISION DETAIL
  https://phabricator.kde.org/D26338

To: esjeon, #plasma, davidedmundson
Cc: asturmlechner, plasma-devel, LeGast00n, The-Feren-OS-Dev, jraleigh, zachus, 
fbampaloukas, GB_2, ragreen, ZrenBot, ngraham, alexeymin, himcesjf, lesliezhai, 
ali-mohamed, jensreuterberg, abetts, sebas, apol, ahiemstra, mart

Reply via email to