[EMAIL PROTECTED] writes:
> We don't keep atime (we dare not make it remotely accurate with the
> amount of caching that's going on), but mtime and ctime should be
> reasonably accurate.
>
Paul Howell <[EMAIL PROTECTED]> writes:
> ctime isn't kept either as near as I can tell.
At MIT, changes were made several years ago to allow the ctime to be
reported back to the client accurately based on the serverdate. (The
patches were submitted to Transarc back then). I suspect the
modifications were never integrated into the distribution and that the
atime, ctime and mtime are still being set to the same value.
(afs_vnodeops.c: afs_CopyOutAttrs).
Ezra
