December 4th, 2005


desktop conference and a bit of a rant..

Just a small follow up to Dave Jones piece on the OSDL Desktop Architecture meeting, notes by Dan.

As far as I know no kernel developers were invited to the meeting, but I think the role of kernel developers particularly in the graphics driver arena isn't a major one, as we don't do much in the kernel, and we've all agreed at KS and DDC last year on the way this thing should go forward. Other subsystems such as wireless, input etc.. would be of more value alright.

The two long standing areas are as usual, memory management and mode setting, Two reasons I can see it hasn't gone forward are:

1. the work isn't sexy cool flashy whatever, its not going to get a press release or a cool demo.
2. a lot of the work is heavy lifting and as such very difficult to throw some newbie or graduate engineer onto, the people who can do this stuff are known, they just happen to be doing other things.

There is also a lot of work on the new X on GL going on my behind closed doors without any community involvement at a certain company who used to be very closed-source based, and seems to think this is the way forward, so they can do a flashy marketed heavy press release. It is impossible to know the scope of the work that company is doing, and I believe they are going to codedrop the whole lot onto without anyone else getting to take a look at the design and review the system. For the future of graphics on Linux I'd rather we blocked this short term thinking in favour of some more involvement with the people who do develop their code in the open. (You know who you are ppl...), I really would like to demo a native Xegl server running on r300 hardware at LCA, however the last codedrop of Xgl is 6 mths old and I know the latest code is a severe improvement on this, I've nearly gotten the r300 driver to a state where it can run Xegl, but I'm going to have to do a lot of working to getting GLX support working to do a cool demo, however I know this work has been done already but the code isn't being released.

BTW if anyone is at LCA I'll be presenting an updated Future of Linux graphics paper which should cover the current status as of the day before I present it (when I write the slides)....