gnu-arch-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnu-arch-users] extended attributes


From: Tom Lord
Subject: Re: [Gnu-arch-users] extended attributes
Date: Tue, 6 Jan 2004 10:11:30 -0800 (PST)

    > From: Andrew Suffield <address@hidden>


    > Note that tools such as cp(1) and mv(1) have the expected
    > behaviour for all of [linux filesystems having extended
    > attributes]. 

"expected behavior?"   Certainly I would expect ACLs to be handled in
a certain way but what is the expected behavior for an arbitrary
user attribute?   One suggested use for user attributes is to record
"coding system".   Must every program that O_CREATs a file be taught
about that attribute and modified to use it appropriately?   

One type program that O_CREATs is a shell as when creating a file to
which the output of some other program has been redirected.  While
`cp' may do something you decide to call reasonable, what happens with
its close cousin:

        cat foo.txt | sed -e 's/Gnu/GNU/' > foo2.txt

(At least ACLs, in that case, can be given a reasonable default for
foo2.txt -- arbitrary user attributes can not.)

In general: the existence of EAs damage the similarity of streams and
files.   They break the unix model.


-t




reply via email to

[Prev in Thread] Current Thread [Next in Thread]