On Fri, Mar 13, 2009 at 2:09 PM, Matt Benjamin <[email protected]> wrote:
> I sense we don't want to ACL info (whether or not per-file) in
> AFSFetchStatus64...
>
> Since you bring that up, what the cache managers do now to cache access
> control information (axsache) doesn't rely on the specific access list
> that might be on an object, right?  A cache manager can get this
> information, but it does so only to permit ACLs to be displayed and
> edited.  So when we specified this aspect of per-file ACLs for GSOC, we
> were assuming that notionally doesn't change, just ACLs may be
> associated with file objects as well.  So clarifying, are you (Derrick)
> thinking of a different model for client caching of access control
> information than we currently have?

not necessarily; i assume if we go ahead with per-file ACLs as was
worked on for GSoC last summer, that we should at least consider
before minting a new RPC if there are any optimizations we'd want
which would need something accomodated in that new RPC.
_______________________________________________
OpenAFS-devel mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-devel

Reply via email to