I just noticed I have an old patch lying around for file.c that implements 
(fully, I think) support for the sub-second fields for access/modification/
create times on files as per the stat(2) structure.  Is this desirable?

In the same vein, I'm thinking the standard Time object should be converted 
to use timespec, not timeval, to allow for no loss of precision when storing 
times with as little granularity as the system allows for.  Note that in the 
patch for file.c I have now, there's a lot of / 1000 being doing on 
nanosecond values to convert to microsecond values :)

What do you all think?

-- 
 Brian Fundakowski Feldman           \  FreeBSD: The Power to Serve!  /
 green / FreeBSD.org                    `------------------------------'