Maybe rsync could support extattrs on non-supporting fs's with a .<fn>.xattr file stored at the same place as the <fn>: If a user used a switch to emulate 'xattrs', then on a fs that didn't support real .xattrs one could still ACL and other .xattr info. If a user ran rsync w/o that switch such files would be treated as 'normal files' -- allowing copying as normal files between other targets, but allowing ACL restoration on some final target by re using the --emulate switch, where it would detect the files on the source, and support on a target, and set ACL's appropriately? Just a thought....? Doable or reasonable? -linda
On Mon, 2009-08-24 at 09:43 -0700, Linda A. Walsh wrote:> Maybe rsync could support extattrs on non-supporting fs's with a > .<fn>.xattr file stored > at the same place as the <fn>: > If a user used a switch to emulate 'xattrs', then on a fs that > didn't support real .xattrs one > could still ACL and other .xattr info. If a user ran rsync w/o that > switch such files > would be treated as 'normal files' -- allowing copying as normal files > between other > targets, but allowing ACL restoration on some final target by re using > the --emulate > switch, where it would detect the files on the source, and support on a > target, and > set ACL's appropriately? > > Just a thought....? Doable or reasonable?I think the xattr emulation would be more appropriately done in a virtual filesystem to be layered over the real one (e.g., with FUSE) than in rsync. -- Matt
Reasonably Related Threads
- wbinfo -i returns the same id for all users, authentication doesn't seem to go through winbind at all
- flac -- exhaustive model search vs. -A <*>?
- BUG: Standard Time v. DST calculated wrong by Samba Server
- Win95, sharing printers and drives
- stepwise selection cox model