Some Desktop Thoughts
After thinking a bit about what is the status of desktop things, and the work desktop team (or Interface Team as it is called) is doing currently I've got some ideas on things we could be doing. It is not a roadmap nor a TODO list, it is just some ideas on how we could improve the user's desktop experience (as we are known to be a desktop-friendly distro, we should think and try to innovate as much as we can on that subject). The topics presented here are just an overview of my thoughts on that. I will try to specify more each idea.
Composite is the way to go
Besides some composite implementations not being much stable, I see the Composite currently as being the way to innovate on desktop. It provides a way for us to composite windows and get real transparency. It also provides an ARGB visual that could be used by the applications to create windows in different formats, and using this real alpha channel to draw alpha-blended windows.
This screenshot shows a simple example of things we can do using composite. KDE4 will for sure be planned to make extensive use of composite (QT4 even uses the ARGB layer if it is available according to Zack Rusin comments).
Composite is not just eye-candy. It can reduce some CPU usage by avoiding redrawing window contents, so we do need to think more about using it.
Some ideas on composite...
What about OpenGL compositing?
That is even cooler. We know the 3D acceleration support on drivers is not the wonder we would like it to be, but we could still use the drivers to start thinking about things that could be done using this brand new concept. When we decided go for X.org 7.1 we've got AIGLX for free. By having AIGLX, we can start doing some GL compositing (compiz, glxcompmgr, etc). We have even Xgl that could be used in cases AIGLX doesn't fit. The problem of GL compositing is that we can't avoid relying on proprietary drivers (unless we target old radeon, intel cards and other not-so-popular chipsets). This is not good as we can't really assure stability and quality. For the open source drivers, they need some work, but at least we can do the work.
Some ideas on OpenGL composite...
Semantic Desktop
I'm not much into this project, but my feeling is that we could communicate more to discuss ideas and implementation. Not that I think guys that are working on this are not good on that, it is that I think we could do even better if we got to discuss things. Desktop search is important in this context of building a semantic desktop, but just if it is well integrated with the environment around. Talking about integration, in order to make the desktop better for the users, it is absolutelly necessary that the main applications are well integrated and have similar behavior. We know this is not completelly true (in KDE the applications share the same look-and-fell, but still they aren't completelly integrated. There is still work to be done). Our custom tools (drakx tools, etc) should also be well integrated in the environment.
Some ideas on Semantic Desktop...
Improving the current desktop
I know we have short time for Mandriva Linux 2007, but I think there is still time for some minor improvements (bugfixing, translation, rework some UIs, etc). Those fixes are not really revolutionary, but they do make difference for the user (who will feel more confident with the desktop if he/she feels it is stable and well prepared).
The future
I'm really glad we are already working on KDE4. Working on the X.org (as we will start doing in short time) is good too. As far as I know Laurent is helping on the QT4 port, and Glenn is working on Solid. Both things are really good. Solid is very important for KDE4, as it will improve the hardware handling and make it transparent for applications (and more portable too). Xegl is the target I'll be working as a long term solution. Still, most of the work done for Xegl could be even used by X.org, so some things may start to appear in the meantime.