I think the basic problem is that 2.0 was released to soon and the
last code freeze was not really a code freeze. Its always been my
experience that changing a *single* line of code in a production
release requires careful consideration and plenty of beta testing.
Linux development operates in a different manner so Linus and others cant
really do that but maybe we could put together a test suite - done by
different beta testers each with there different equipment and setups. We
could design the tests to test each of the kernel subsystems. New versions
could be marked beta until they pass all the tests. This wont catch all
problems but it seems some of the problems since the last 1.99.x kernel
could have been caught this way. I know no one has the time to
organize such a scheme but if ever they do I will be willing to beta test
because I tend to run the latest kernel on my LAN at home. Anyway, it looks
like the big distibution makers are not pressing their new 2.0 based CDs
yet so we wont end up with bad versions of distributions floating around.
This is important because its easy for me to tell someone "get the
latest xxx CD and just install" rather than "get the latest xxx CD and
install and then download kernel patch xxx and compile and dont run xxx
until thats been done" , when they hear the latter they tend to want to wait.