Re: Content negotiation

Kee Hinckley (nazgul@utopia.com)
Wed, 8 Nov 1995 01:36:12 -0500


At 4:34 PM 11/7/95, Eric W. Sink wrote:
>> Spyglass_Mosaic/2.10 Win32 Open Text/1
>> PATHWORKS Mosaic/1.0 libwww/2.15_Spyglass
>> Mosaic/1.0 libwww/2.15_Spyglass
>> Enhanced_Mosaic/2.00 Win32 FTP Software/Spyglass/3
>
...
>string format has changed a bit. The third example above is *really*
>old! :-) Anyway, we have 46 licensees. Coming up with a user-agent

Just goes to show. We got that hit on November 2nd, from Norway. No
matter how hard you try, you can't get rid of those old versions :-).

>FYI, Currently, User-Agent strings for all builds done by Spyglass
>follow this convention:
>
>Spyglass_Mosaic/BASELINEVERSION PLATFORM VENDOR/BUILDNUMBER

Great, that's helpful.

>> As a content
>> developer, the most I hope for is that companies like Spyglass try and ma=
ke
>> the task easier, not harder.
>
>Great! We'd be delighted to help. How can we help make real content
>negotiation a viable option for you?

The two main issues I've mentioned already. One is HTML/HTTP extensions.
Obviously you can't spell out every single on in the headers, but some
attempt to clump and label them might work (e.g. HTML2.0+tables). Probably
more important though is the issue of supported content-types. We are
doing a couple sites right now that have heavy audio content. Ideally
we'll serve up RealAudio in real time, but we have to offer alternatives,
and the only way to make it really work is to ask the user up-front when
they visit the Web Site. "Which audio-formats can you listen too..."
That's not a solution I enjoy, and if we even if we are very careful with
the wording, it's one that some users will answer incorrectly. I haven't
been tracking the standards efforts in the last few months (keeping up with
the non-standard efforts is a fulltime job), so I don't know what's going
on in this space. But fundamentally I believe you've got to have the
browser send a list of all supported content-types (and of course, send a
different list for inline GETs than for external ones). Unfortunately it
means added overhead, probably less than 1K, but still, that's another
second for slow connections. Registered content-types could be encoded
for efficiency I suppose, but that would mean changing the syntax of the
accept field. The point of HTML is to let the browser mediate the
presentation, not to force the user to do it.

Kee Hinckley Utopia Inc. - Cyberspace Architects=81 617.768.5500
nazgul@utopia.com http://www.utopia.com/

I'm not sure which upsets me more: that people are so unwilling to accept
responsibility for their own actions, or that they are so eager to regulate
everyone else's.