On Tue, 20 Feb 2001, Stephen Tweedie wrote:

> Hi,
> 
> On Tue, Feb 20, 2001 at 03:51:18PM +0100, Andreas Gruenbacher wrote:
> > > 
> > > This still seems to offer no way of solving the big problem I've been
> > > trying to stress for months now: you have namespace problems *inside*
> > > an ACL namespace.  
> > 
> > Yes. The proposal tries to address what's an extended attribute, not
> > what's inside an extended attribute. Even that proved pretty complex.
> > There is no mention of what's inside any namespace, or what's inside a
> > value. It only says there are additional restrictions on the permissible
> > values for some namespaces.
> > 
> > I see a couple of your ideas fit perfectly in a "what's inside an extended
> > attribute" specification, though you seem to believe they're more of a
> > "what extra operations are in a namespace" issue.
> 
> We're going round in circles again. :-)
> 
> Why should the ACL API have to assume that the ACLs are in fact stored
> inside EAs?  It is perfectly valid for a filesystem to support ACLs
> without supporting generic EAs.

Then let's assume an ACL syscall, if you wish. The problem stays exactly
the same. You have to have a representation for ACLs that you can pass to
the kernel (in one piece). This representation obviously needs to be
flexible enough to cover various types of ACLs, and if it is by saying
"this is a POSIX ACL with numeric IDs".

I've never said ACLs must or even should be stored as EAs---they could be
stored anywhere. They are just passed to the kernel via an interface that
happens also to be used for extended attributes, so to say, because this
interface fits ok.

Most filesystems won't have a dedicated ACL store. Also, they will most
probably not want to store ACLs in this most general form, so some
conversion will have to take place. Also, the values of ACLs need to be
checked. There has to be some verification mechanism.

Then, the filesystem can easily store the ACL as an extended attribute,
just as easily as it can store it differently. The filesystem knows it's
dealing with an ACL, nows whether it is acceptable, and can decide to do
with it whatever it wants.

I see no problem here. Where do you?


Cheers,
Andreas.

------------------------------------------------------------------------
 Andreas Gruenbacher, [EMAIL PROTECTED]
 Contact information: http://www.bestbits.at/~ag/

-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]

Reply via email to