> Well, while I understand that having a common com code for all possible
> architectures is a good idea, I'm not sure if it's a good idea to reject
> an i386 com driver which clearly works better than the one we have now.

It's a bloody well _GREAT_ idea, if somebody tries to put said driver
in sys/dev/isa, where it's NOT just an 'i386' com driver.

If you're going to call it machine-independent, it should be
machine-independent.  Currently, the 'com' driver is.  if somebody
wants to have an i386-specific serial driver, they can easily do that.
I see no reason the driver can't work well on all architectures.
given that, i have to think that people who want to make it
machine-specific to solve short-term bugs are short-sighted, at best.


cgd