umm, as far as i can tell, every UN*X that i have access to (net/2
   derivatives, 4.4BSD, Ultrix, SunOS, Domain/OS) does exactly this.
   it's the expected behaviour, not a bug.  RCS is in error here, and the
   vfs routines should not be "fixed."

I didn't have a chance to check this behavior on other systems & I
probably should have.  Feel free to cancel out this bug.

I have to admit, having looked at the code in RCS in question, fixing
the kernel was easier (I do *not* want to turn off the -DFSYNC_ALL
option for what should be obvious reasons).

						- Bill

------------------------------------------------------------------------------