On Tue, 16 Feb 93 15:12:21 -0500 John F Carr wrote:
>
>I think the top level of AFS at a site should only have local directories.
>There should be a subdirectory of /afs where mountpoints to remote cells are
>kept. This has a couple advantages: (1) "ls /afs" is fast and does not
>cause the machine to become aware of remote cells (for every remote cell the
>machine knows about, there is net traffic for callback maintainance), (2)
>pathnames to local files are shorter.
>
If you do this, then sharing files between cells becomes a nightmare.
If a shell script is run in cell A, its path is /afs/A/xxx while if you
run it from cell B, its path needs to be /afs/foreign/A/xxx. And of
course, someone will want to spell foreign as foriegn or Foreign ... I
don't like that can of worms.
,,,
(o o)
-------------------------------+-----------------oOO--(_)--OOo----------
Frank Swasey | IBMIPNET: [EMAIL PROTECTED]
I&TSS NorthEast Region | Internet: [EMAIL PROTECTED]
Distributed Computing Services | BITNET: FSWASEY at VNET
D54/615-1 | VNET: FSWASEY at BTV
(802) 871-4011 (tie 435) | fax: (802) 871-4253 (tie 435)
"The most important political office is that of private citizen."
-- Louis Brandeis
