Re: SPEC: DNS conventions & "naming" of VRML sites

The protocol has outgrown its name and original intent as a hypertext
'only' transfer protocol (as has the hypertext markup langauge, which is
now really a hypermedia format since you can inline images).

IMHO a Media Transfer Protocol (MTP), as it i negotiation (of type, langu, etc.) and transportation of any
digital media (not good support for contina>

] Perhaps what we should be looking to is a vrtp:// protocol which would
] contain gzip compression as well...

No, please.... The protocol has nothing to do with the contents
it transports; it is on a lower layer. When VRML behaviour i to be addressed, however, the protocol will need to chanb>; either it
will become a completely VR-specific protocol, or HTTP will rise to the
occasion.

Compression is supported by todays standards. With CERN httpd 3.0 and
Netscape 1.1 at least, if you try to access a URL that is mapped to
filename.html on the server's filesystem, and the server only finds
filename.html.gz, the server will send the compressed file and set
content-encoding to x-gzip. Netscape understands how to decode
x-gzip and does that before handling the content.

About using vrml.bla.bla instead of www.bla.bla, it really doesn't
matter. :) But, (again IMHO) www.bla.bla may be better because it
reflects that you conntect to a www server that talks HTTP, it doesn't
imply anything about the content types the server serves, and is thus
more robust to format/content-type chanb>s.

Anyhow, I wouldn't want anyone to use vrml.bla.bla just because
they think or else they may end up chanbing it frequently and chanbing
names/URLs is a big maintenance problem on the WWW today.

Bye,

-- 
Bjoern Stabell <mailto::[email protected]>
               <http://www.cc.uit.no/~bjoerns/>