On Thu, Apr 04, 2002 at 01:40:31AM +0300, Jarkko Hietaniemi wrote: > On Wed, Apr 03, 2002 at 05:37:00PM -0500, Michael G Schwern wrote: > > On Wed, Apr 03, 2002 at 11:37:21PM +0300, Jarkko Hietaniemi wrote: > > > NFS likely. I don't know what to do about this, detecting > > > that we are on NFS is ENONPORTABLE. > > > > Touch a file and see if it's mtime is significantly different than the > > current time? :) > > That gives you "we *might* be on NFS", when the situation might be > any of the "we are on working NFS" (oxymoron?), "we are on something > else broken", "your clock is broken", or something else...
If you can execute `df .` is it portable enough to assume that if a line matches m!^\S+:/! then you are on NFS. Or it's probable enough that you can issue a warning that you suspect you're on NFS. If you can't run df, or you find no such line, you don't say anything. [rather than saying that you aren't on NFS] But something is telling me that it's not possible in a portable way to suppress error messages about df not being found. And now is not the time to start adding probes to Configure for df. Nicholas Clark
