From: Spencer Shepler (shepler@eng.sun.com)
Date: 07/17/02-05:53:51 PM Z
Date: Wed, 17 Jul 2002 17:53:51 -0500 From: Spencer Shepler <shepler@eng.sun.com> Subject: Re: subset of ACLs functionality Message-ID: <20020717225351.GG100440@shepler.eng.sun.com> On Wed, Noveck, Dave wrote: > > I don't see any statement in the spec that the same credentials have > to be used on a READ or WRITE as were used on the OPEN. I would expect > that in most cases that would be true but it's not guaranteed, and > therefore you had better check when the IO is done. True and as you point out most clients will use the same credentials or they won't get what they expect. > Even if there were such a rule and the server checked tha the credentials > are the same, there is still the issue of responding to permission > changes. Waiting unitl close and a new OPEN might not be desirable. >From a Unix perspective, check at OPEN is exactly what is desired, right? We are all familiar with the old applications that like to open() then chmod such that the file can not be read/written and this works for v2/v3 when the credential matches the owner of the file but what you have suggested may potentially break then unless the v4 server implements the same behavior as in v2/v3. Spencer
This archive was generated by hypermail 2.1.2 : 03/04/05-01:49:57 AM Z CST