On Thursday, February 27, 2003, at 07:56 AM, Greg Ames wrote:
Most of us have committed bug fixes with the best of intentions which were not quite complete or had unintended side effects. I certainly have - the deferred write pool stuff in core_output_filter comes to mind. Letting the fixes age a bit in the unstable tree reduces the probability of unpleasant surprises happening in the stable tree, at least for mainline code. We can be extra diligent about reviewing/testing changes that we know are not mainline.
I see the need for letting patches age in the unstable tree, but I think we could do that without having to vote on each and every change.
Yeah, that's what I meant by "lazy consensus" in the STATUS file.
But I can also see problems with that being the policy. What's mainline code vs. non-mainline? How long is long enough? The answer could change around Hannukkah/Christmas/New Year's LinuxWorld/ApacheCon/whatever.
Greg
