From: Srinivas Bharadwaj (sbharadw@concentric.net)
Date: 03/12/97-12:55:42 PM Z
Message-Id: <3.0.32.19970312105541.0068b178@smtp.concentric.net> Date: Wed, 12 Mar 1997 10:55:42 -0800 From: Srinivas Bharadwaj <sbharadw@concentric.net> Subject: Re: Attributes > >Some starting criteria -- > >It solves a *real* problem that exists for customers today. "Customers" should be anybody with a computer and a network connection. (a modem, LAN connection etc.) >It has a champion to explain the intended semantics. A must. >It is implementable in more than one real operating system, including at > least UNIX and probably NT or Windows or some such. Yes, in general they should not be too OS specific. It is necessary that complicated features should be optional or relegated to minor versions(generalized ACLs for example). >It should also be implementable by a Web client. Not for optional features. >Any client should able to interoperate with any server. For the overall protocol, but not for the specific feature being proposed. (minor version 5 might be best used by Web clients(may be meaningless on a UNIX machine on a LAN) - a web client tries minor version 5 and if that fails he goes back to V4 - minor version 3 or maybe even V3). Precedents exist for this, NFS V3 had the notion of NFS3_NOTSUP. Like NFS V3 there should be a base that is mandatory(or may be V3 can be that base) that will allow any client to operate with any server, but not every feature. > >> >>I would ask people to be aware that this is a "file access protocol", >>intended for use by more than just Web/Java applications and is real, >>to be used in production environments. >> >> Thanx... >> >> ps >> >> srini
This archive was generated by hypermail 2.1.2 : 03/04/05-01:45:31 AM Z CST