>>hahahha you are on the root-replicated volume....
>>What are you doing here .????? :o)
>>
>:(
>

This file is here to remember be if I wrongly chroot.
That's not a good argument i've used it for a test..ok, ok...
No matter it's my test machine !
Concerning acl behaviour have a look at :
http://www.coda.cs.cmu.edu/doc/html/kernel-venus-protocol-4.html
Shame to me ! Really time to read docs...
All fs-primitives have been rewrite in coda. How could it be different ?

>Yes, this is like in AFS. But what is with the token? Must I type in my
>password two times, one for kerberos and one for coda?
>  
>

Again I've never used kerberos....
Correct me if wrong but it's just an auth server used at identification 
time ( ftp, local login...etc)..
Thinking on this I don't see any reason to such a behaviour !
It would not be so technicaly impossible to use the STDIN password and 
send it to clog, if auth-call vs kerberos return is something like " 
auth granted"...  After all isn't that the purpose of the pam_kerberos ?
Mmmmm..... but where is the STDIN ?

Or perhaps the behave is  something like :
    login & pass ---------> kerberos
                                               ||
PAM kerberos <--------- chekvalidity
if (return == "ok"){
     system (clog login_at_coda_cell);
}

I don't know how Mr Popov implemented it, and I don't know kerberos 
mecanism.
I should stop speaking of things I don't know....

Think you'd better see the code, try, or ask him.
And I should  stop thinking, exciting,writing stupid things....
and start working...

Endly I don't see any reasons uid to be the same...  kerberos required ?
The only one i see would be for simplification reasons...

-- 
Lionix
FS-Realm (newbee?) Administrator
Hundreds hours of work but so powerful !
And still have to work