begin quoting David Brown as of Tue, Dec 04, 2007 at 11:22:23PM -0800: > On Tue, Dec 04, 2007 at 10:47:49PM -0800, Andrew Lentvorski wrote: > > >First, the Linux kernel has been doing more than little bit of wheel > >spinning in the 2.6.X series. An actual tracking system along with a > >testing methodology would prevent quite a bit of that. > > Very odd comments. There is a rediculous amount of work being done on the
s/red/rid/ > 2.6.x kernels, and not just minor stuff. It's a kernel, though, lots of > stuff isn't visible, it just gets better, or supports more things. > > Let's see: > > % git diff --stat v2.6.12..v2.6.23.9 | tail -1 > 21938 files changed, 3644256 insertions(+), 1854755 deletions(-) > > I'm not sure I'd call 3.6 million lines of code to be wheel spinning. Um... Wheel spinning is a lot of effort and not much progress. 3.6 million lines of code without much to show for it is a pretty good example of wheel-spinning; it would be better if those 3.6 million lines of code fixed 300,000 bugs, say. Doing so much in the kernel does serious violence to my personal aesthetic of code. > They also have a fairly effective testing methodology--although a bit > unusual. Send patches and let thousands of people try them. By that metric, so does Microsoft. And we tend to criticize 'em for it. > I think the kernel is an outstanding example of how, at least in the right > circumstances, a very non-traditional development model can work quite > well. It has it's disadvantages, especially if what someone ones doesn't > fit the kernel developers priorities. It's not how well or poorly the bear dances... -- How many of the other projects set up like the kernel have succeeded? Stewart Stremler -- KPLUG-List@kernel-panic.org http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
