JUL
12
2006
|
On KWin wobbling and such stuffAs some might have noticed, KWin is supposed to get compositing support, allowing a wider range of various effects and replacing KDE3.x's separate kompmgr (developed by Thomas Lübking, based on the original xcompmgr, and according to e.g. this doing rather well for its time). This is in line with today's general belief that compositing managers should not be separate but part of window managers, as it allows for example better syncronization of effects and their wider range (and it will also allow us to get rid of the plain C that kompmgr is, bleh). What happened was that at the KDE four core meeting I had some trouble with my laptop and before Trolltech people managed to bring me another machine to use (thanks!) my possibilities were somewhat limited to things like talking to people (like to Zack about compositing managers), reading text files (like thoroughly reading docs about the new X extensions that compositing managers use or sources of already existing compositing managers) or thinking (like thinking about how much KWin's design would be affected by adding compositing support). Surprisingly it turned out mostly looking like rather simple thing to do and I actually wonder why there hasn't been already a small army of eager graphic coders trying to add that to KWin sooner. Anyway, it apparently ended up being me, so I tried it. Using Fredrik's great compositing manager (that was originally meant as kompmgr replacement for KDE4) was big help. Now, to stop all the cheering, there's one problem. For officially being a GUI developer I'm awfully lame when it comes to graphics. KWin in the kwin_composite branch should have better potential than kompmgr, but right now it can't even do as much as kompmgr. The plan was that the graphics part would be done by Zack, because, as somebody mentioned, he appears to be very talented in that area, however it seems the problem there is that he appears to be very talented in way too many areas, so I guess I rather shouldn't count on him finding time for this. So plan B is to find that army of graphic coders and make it come out of hiding. It looks like I've already found one member, but more wouldn't hurt. I can do all the other parts like handling of effect plugins, tracking window damage and so on, but I apparently can't do the graphics part. So if you know OpenGL or XRender and want to see something more than just plain transparency and shadows in KDE4, -> [email protected]. KWin seems to be in need of somebody who will design the system for graphical effects and code it.
|
![]() |
Comments
Helping spreading news about this
Hi,
I hope you don't mind if I help spread news about this need for more developers that could work on this. I submitted a story with the link to this blog entry to digg.com:
KDE lookking for people to bring more eyecandy to KDE 4
I hope the story is OK and I hope it gets enough diggs to come to the front page.
I'll also post a note about this on my blog.
--
Live long and prosper!
JLP's Blog - http://jlp.holodeck1.com/blog/
XGL/AIGLX
Hi. This sounds very cool!
Will the new kwin be able to accelerate compositing via XGL and/or AIGLX like compiz?
The plan is to have backends
The plan is to have backends for plain xrender as well as for opengl-based x-servers. If course the more advanced effects will only work with opengl.
HTH,
Daniel
Looking at existing projects
Hi, nice that something happens on the kde-front.
Some thoughts: It might be a good idea to consider re-using already existing stuff. That is for one libcm (which is currently used by metacity) and compiz.
libcm is already an abstraction for effects in a separate library. Though I don't know how much work it would be to adopt it to kwin.
compiz on the other side is definitely the most advanced project at the moment, it has much better performance than libcm and much more effects. It is very modular and flexible, e.g. it's designed to have window-decorations in a separate executable. The kde-window-decorator is not working atm (afaik initially done by zack rusin), maybe it would be a good move to get this back running.
Using compiz would mean completely replacing kwin, which is probably a very big change for a project like kde, but imho it should be considered. Due to it's modular structure, compiz is very customizable and could be made completely kde-ish when adding a kde-config-plugin and a good window-decorator.
In opinion of my friends who
In opinion of my friends who use KDE 4 and imho, this version is more convenient and better. One of the main strong points of introduced alterations is improvement of window manager.