SIGs/KDE/KDMtoLightDM

From FedoraProject

< SIGs | KDE(Difference between revisions)
Jump to: navigation, search
 
Line 1: Line 1:
=Switching from KDM to LightDM login manager=
+
=Finding a KDM substitution for future Fedora KDE releases=
  
As it is possible to compile <code>kde-workspace</code> without KDM using a cmake parameter [https://git.reviewboard.kde.org/r/106444/diff/#index_header], we can replace the default KDE login manager (KDM) with others, namely LightDM that has improved its KDE support dramatically lately.
+
== Rationale ==
 +
KDM is obsolete (still bears the XDM heritage) and nearly unmaintainable.
  
== Pro: ==
+
KDM is buggy.
  
* Code base - more recent, supports multiple graphical backends as Ubuntu just announced Mir (so adding Wayland eventually will be easier)
+
KDM won't survive switch to KDE Plasma Workspaces 2.
** In comparison, KDM contains major portion of code of XDM which means the backend is basically pure C communicating with the C++/Qt frontend. Many very handy facilities for managing X are missing, not mentioning the fact there is going to be need to support Wayland in future.
+
* Better performance (according to their presentation, my experince so far is not that great - [[User:Mbriza|mbriza]]) and "light"
+
** I'm a bit sceptical about this point, will measure precise memory and CPU requirements in the future..
+
* Simpler API for creating greeters
+
* Powermanagement is present
+
** Although it is present, is it really a pro as we'll be adding an other powermanager separated from PowerDevil? The settings would differ or would need to be propagated. Also, it's now provided by systemd, too. Will have to investigate further.
+
* Multi-head support is better - Prompt is displayed on the screen where the mouse cursor is
+
  
== Con: ==
+
KDM is written in C despite object orientation of most parts of its design.
  
* solved: Switching users in KDE is not possible. [https://bugzilla.redhat.com/show_bug.cgi?id=859347]
+
KDM doesn't support QML/QtQuick themes.
* Dependency on glib
+
 
 +
We can get rid of KDM entirely, there's a cmake switch for that in kde-workspace: [https://git.reviewboard.kde.org/r/106444/diff/#index_header]
 +
 
 +
== What can we do now ==
 +
=== Further maintenance of KDM ===
 +
KDM doesn't have any really active maintainer as of July 2013.
 +
 
 +
Many bugfixes (not even mentioning features) would require a major structure change and rewrites in many parts of the project.
 +
 
 +
=== Switch to SDDM ===
 +
New young project with decreasing activity of the main developer.
 +
 
 +
Those of you who talked to me about it already know I'm convinced SDDM is the way to go. When time allows, I'm trying to add XDMCP support to it and have promised to write and test logind-style multi-seat, too. --mbriza
 +
 
 +
Pros:
 +
* pure C++ and Qt (already prepared for Qt5)
 +
* nice, clean and logical design
 +
 
 +
Cons:
 +
* Many missing features (most notable: XDMCP, VNC)
 +
* Decreasing developer activity (which was really huge initially, reduced pace was quite expectable)
 +
 
 +
=== Switch to LightDM ===
 +
The main DM of Ubuntu is developed by Robert Ancell and is covered by the Canonical CLA.
 +
 
 +
As Ubuntu will be switching to Mir (in 13.10 as far as I remember), we'll see how they'll handle using it instead of X as a backend. If it'll be possible to easily switch those backends, there would be a ground prepared to add a Wayland one.
 +
 
 +
Pros:
 +
* David Edmundson's Qt bindings for QML/QtQuick greeters
 +
* Provides API for creating greeters in virtually any form
 +
* Actively maintained and developed
 +
 
 +
Cons:
 +
* Written in C using GLIB
 +
* CLA
 +
* To me, it really doesn't seem "light" enough, starts slower than KDM and takes more memory, too.
  
 
== What needs to be done: ==
 
== What needs to be done: ==

Latest revision as of 16:41, 4 July 2013

Contents

[edit] Finding a KDM substitution for future Fedora KDE releases

[edit] Rationale

KDM is obsolete (still bears the XDM heritage) and nearly unmaintainable.

KDM is buggy.

KDM won't survive switch to KDE Plasma Workspaces 2.

KDM is written in C despite object orientation of most parts of its design.

KDM doesn't support QML/QtQuick themes.

We can get rid of KDM entirely, there's a cmake switch for that in kde-workspace: [1]

[edit] What can we do now

[edit] Further maintenance of KDM

KDM doesn't have any really active maintainer as of July 2013.

Many bugfixes (not even mentioning features) would require a major structure change and rewrites in many parts of the project.

[edit] Switch to SDDM

New young project with decreasing activity of the main developer.

Those of you who talked to me about it already know I'm convinced SDDM is the way to go. When time allows, I'm trying to add XDMCP support to it and have promised to write and test logind-style multi-seat, too. --mbriza

Pros:

  • pure C++ and Qt (already prepared for Qt5)
  • nice, clean and logical design

Cons:

  • Many missing features (most notable: XDMCP, VNC)
  • Decreasing developer activity (which was really huge initially, reduced pace was quite expectable)

[edit] Switch to LightDM

The main DM of Ubuntu is developed by Robert Ancell and is covered by the Canonical CLA.

As Ubuntu will be switching to Mir (in 13.10 as far as I remember), we'll see how they'll handle using it instead of X as a backend. If it'll be possible to easily switch those backends, there would be a ground prepared to add a Wayland one.

Pros:

  • David Edmundson's Qt bindings for QML/QtQuick greeters
  • Provides API for creating greeters in virtually any form
  • Actively maintained and developed

Cons:

  • Written in C using GLIB
  • CLA
  • To me, it really doesn't seem "light" enough, starts slower than KDM and takes more memory, too.

[edit] What needs to be done:

  • Add the conditional KDM inclusion parameter to kde-workspace
  • Edit kde-settings not to ship KDM settings

[edit] Links