From: Spencer Shepler (shepler@eng.sun.com)
Date: 07/23/01-06:01:29 PM Z
Date: Mon, 23 Jul 2001 18:01:29 -0500 From: Spencer Shepler <shepler@eng.sun.com> Subject: Re: "recommended" attributes Message-ID: <20010723180129.J519@dhcp-aus08-189.eng.sun.com> I agree on both counts (original intent and plan of action). As owner of the ever growing (and tardy) issues list for RFC3010, it seems that we should limit ourselves to what is "broken" or needs clarification. Minor versioning will allow the introduction of additional attribute definitions and other functionality. It would be good to get some operational experience with NFSv4 before jumping into adding addition options... On Mon, Noveck, Dave wrote: > It seems to me that recommended attributes started out > as just that: attributes that we recommended that > servers implement but which we recognized that not > all servers might be able to implement. The idea > was that it was better for servers to honestly tells us > that no support was available, than to try to pretend > (badly) that support was available. > > At some point, we started to add attributes of which > that could not be said (e.g mime-type). The newly- > proposed attributes also do not seem to be things that > we are not intending to really recommend, but instead > are optional attributes, that some servers might > implement and some clients might look at. > > I'm not dead set against such optional attributes, > but we have to recognize that their inclusion > complicates nfs inter-operability testing (not the > part that consists of running the connectathon > tests, however :-). Since we can add recommended/ > optional attributes in a minor version, I think we > should go slow about adding these now. Let's not add > attributes because they exist in some filesystem > and might at some point be needed. If a real need > for an attribute arises, we can add it at that point. > > > -- - Spencer -
This archive was generated by hypermail 2.1.2 : 03/04/05-01:48:59 AM Z CST