>    If you want to change all of the bcopy/bzero instances to mem*
>    equivalents, that's one thing, but don't add new functions to the
>    kernel because somebody screwed up some new code that they wrote.
> 
> It's a *far* stretch to say that someone `screwed up' in this case,
> unless you're referring to the fact that macros in the *kernel* include
> files use functions that aren't present in the kernel.

Actually, there are a whole bunch of screw ups here:

	(1) a change was checked in that caused the kernel, compiled
	    "as expected," not to link.  It's one thing to say
	    "the -current kernel is going to be unstable/won't compile/
	    whatever for a few days," and another to have that happen
	    accidentally.  This is definitely a screw-up, and it makes
	    me wonder how you could possibly have compiled and tested
	    the changes you made.

	(2) (as you note) there was something in a kernel header which
	    shouldn't have been used in the kernel as-is, but was
	    defined for use in the kernel (i.e. not #ifndef _KERNEL),
	    and wasn't documented as "not for use in kernel."

	(3) when the bogon _did_ show up, nothing at all was done
	    about it for several days, probably because people
	    thought that since you'd been actively working on the
	    code, you'd fix it in a timely manner.


None of those are what i'd call "serious" screw-ups, and screw-ups
like that are pretty much to be expected if the source is under
development.

That doesn't meant that they're not screw-ups, though.



cgd