first of all, im an ultra hardcore fan of kde since kde 2

about kde 4.x never crashed for me, about speed well in 4.3 series i can say for me is already fixed, so great work make kde as uber as it is, i have to say that after use kde everything else feels just wrong especially codesaladgnome.

now with the "but" part:

* kwin compose : i have to say lately feels very stable, but i think compiz still have more options and feels a bit more snappy especially in laptop with **** gfx (ofc i agree compiz is less stable now than kwin is). i dont know for sure if this is possible but wouldnt be possible to keep the basic initialization code inside kwin and extract all the actual render code to a modular/plugins system??. i think this way would be faster to optimize/add effects to kwin without touch kde basecode and well even join compiz here so more ppl coding + compiz team maintaining the code. this way any DE project should be able to use higly optimized effect accessing this set of libraries for example?? like phonon and xine beside the obvious less duplication and more devs

The problem with compiz, from what I understand, is that it is based on a very broken and hard-to-maintain code base. This has led to, I think, three or four internal forks of the project right now, and there has been very little visible progress over the last few years. If I recall correclt, early on KDE developers considered the idea but rejected it for these sorts of reasons.

KWin effects are already ( internally at least ) plugin based. I do not know if it possible to add 3rd party effects, without altering the KWin code base directly.

The problem with compiz plugins particularly, as I recall, is that there was really no compiz api, they just connected right into the internals. This was considered by the KDE team to be a stability and security risk and didn't want to expose kwin to that sort of low-level manipulation.

mmm, yes make sense like all coming from gnome or thinked for gnome (the code salad way ). then in that case maybe kde developers can take the initiative to at least start the project and put some widget toolkit neutral C++ code based in the kwin compose system, even maybe stablish an initial api/goals. perhaps more dev will support this system instead of compiz. i need more practice with openmp/pthread (and ofc update my qt knowledge to 4.5) code but i think in some weeks at least i could help working on the qt/kde bindings in that case

I don't know if this any big news or not, but I believe the compiz dev's are working to create a much better codebase and port everything (including all plugins) to C++/QT. By release 1.0 they say it should be much better and work very well within KDE. That's what the dev's say anyway:/ I'd hope they are right. They seem to favour working to make compiz useful inside of KDE, especially since GNOME has completely dumped compiz for GNOME 3 (currently due next year in fall).

that would be sweet paul

jrch2k10 wrote:that would be sweet paul

Well let's just hope the compiz dev's follow through with their plan.

I like Kwin, and it's very smooth & well-integrated. However I like to run compiz for the heck of it, and my little brother loves it when he uses my laptop from time to time. Besides, compiz runs crazy-smooth and fast for my laptop. Even with all kinds of plugins enabled. If the dev's follow through with their plans, 1.0 could be one heck of a release!