On Tue, 2007-05-15 at 02:50 +0000, David Wuertele wrote: > What I found was that the dependency (file being copied by cp -a) had > a non-integral mtime, but the target always had an integer mtime:
> Can anyone explain why cp -a might copy the fractional portion of the mtime > *sometimes* but not others? Your attempt to reproduce the behavior is invalidated by the fact that you didn't duplicate the real conditions: in your test you copied one file to another in the same directory. In the real environment you copied one file to a completely different filesystem, and not only that but it was a different TYPE of filesystem (local, presumably ext3 or ufs or something, vs. NFS). Timestamps are stored on the filesystem and so the maximum resolution for the timestamp is dictated by (among other things) the filesystem type. In this case, apparently your implementation of NFS does not have support for sub-second timestamps, or else cp is behaving differently if it detects that the remote filesystem is NFS. You can look up the .LOW_RESOLUTION_TIMESTAMP pseudo target in the GNU make manual for one way to manage this situation. -- ------------------------------------------------------------------------------- Paul D. Smith <[EMAIL PROTECTED]> Find some GNU make tips at: http://www.gnu.org http://make.paulandlesley.org "Please remain calm...I may be mad, but I am a professional." --Mad Scientist _______________________________________________ Help-make mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-make
