On Mon, 2002-10-28 at 15:30, Jens Owen wrote: > There was a discussion today in the weekly DRI developer's meeting > regarding the cross purposes of pushing the latest DRI infrastructure > work vs. providing a stable, easy to use set of drivers for our broader > base of BETA testers. > > The conclusions reached in this discussion were: > > 1) The DRI infrastructure work is the most important for the long term > success of our project. We should not let any effort to address a > stable driver branch or release interfere with our primary focus of > bringing new functionality to the DRI. > > 2) Point 1 being said, it's unrealistic to think that we can ignore the > fact that there is a huge time lag between when new driver functionality > is available and when it starts to be supported in most distributions. > If we can find a way to release stable driver updates on perhaps an > infrequent basis (quarterly), that would address this huge time lag and > enable many more users access to the current DRI drivers they need. > > 3) Keith agreed to be the unwilling maintainer of the stable driver > branch until another experienced DRI developer volunteers. He will only > be doing minor updates for patches that fix high visibility problems he > feels should be addressed. If you're interested in maintaining the > stable driver branch, send e-mail to dri-devel. > > 4) Reide volunteered to build, package and release the stable driver > branch when it's ready. We probably won't need nightly builds, so the > scripts can be kicked off by hand once the build process is ready and we > have any source updates. Jose volunteered to help Reide getting the > binary driver releases going.
I would certainly be willing to work on maintaining the stable branch with respect to FreeBSD if it's a general "stable" branch, and make packages of it if I find the time to set up that build machine I've been wanting to make (first I have to wait for those replacement drives, ugh). I would prefer if releases (to X, binary packages on the website) were made from the stable branch, but it would require committers to commit twice, once to the current development branch and then to stable later after the patch has been shown to work well. That way most development, even mach64 and such, could occur on the current branch. I wish I could have been at the discussion, but the meeting time with or without daylight savings time conflicts with classes :( -- Eric Anholt <[EMAIL PROTECTED]> http://people.freebsd.org/~anholt/dri/ ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
