> As for Bastien's point, "RTFM", I think that is appropriate for a point
> upgrade,
> e.g. slink to potato, but a minor upgrade to 2.2r2 should really not require
> paging
> through oodles of kernel docs in the console (because X won't start without
> the
> pointer). Maybe we need some kind of call for testing before a new rev
> release, to
> catch problems like this?
Maybe we just do without pulling in the new input device code (which
breaks mouse support in unexpected ways) in a point release? For that
matter, I'd stay away from kernels using ix86 keycodes for console as
well. More breakage without even the slightest attempt of keeping
backwards compatibility. Bad idea.
Michael
