Re: More WebNFS vs CIFS notices

Henrik Frystyk Nielsen (frystyk@w3.org)
Mon, 26 Aug 1996 18:40:58 -0400


Rohit Khare writes:
> This is really pathetic. w3mux *ought* to blow both away and consign them to
> the dustbin of history. But not a peep from W3C or any other web developer to
> stand in the way of a solidifying industry battleground. The theory is, let's
> test and be sure that w3mux is perfect before tipping off the world that these
> protocols are completely the wrong solution... [no caching, proxying,
> application semantics, versioning, etc]

You can get exactly the same behavior in HTTP/1.1 right now using byte ranges. It would be nice to see some statistics about performance between WebNFS and HTTP/1.1 using persistent connections. WebNFS is rather verbose and will not be a lot faster. Especially if we get some of the performance issues solved in HTTP/1.2 then the difference will be very small.

I just checked the sun pages to see the status of webnfs

http://www.sun.com/960710/feature2/index.html

> Industry leaders Oracle Corporation, Spyglass, and JavaSoft plan to
> put SunSoft WebNFS into their browser software, and IBM RS 6000,
> Sequent, Sun and Auspex Systems will add WebNFS capabilities to their
> server software. This wide adoption of WebNFS will strengthen the
> World Wide Web computing infrastructure and pave the way for WebNFS
> to become the standard file access protocol of the Internet.

Either their idea about the Web is kind of strange or else they actually do not claim to have found the replacement of HTTP. Maybe Sun simply has discovered the Internet and that FTP is not a smart solution.

However, it goes on and to my big surprise, I read the following:

> WebNFS server source is now available in the current release of ONC+
> 2.0. WebNFS client source will be made available to the W3C Reference
> Library so that developers can build browsers that include support
> for WebNFS. WebNFS will be submitted for publication by the
> International Engineering Task Force (IETF) as Informational RFCs.
> Vendors will be able to build conforming implementation from these
> specifications.

There we go - a nice link to my toy libwww. Isn't that nice?

Henrik Frystyk