- From: Brian Peterson <brianp@apocalypse.org>
- Date: Sat, 27 Jan 1996 17:00:37 -0500
- To: davido@apocalypse.org, gjw@best.com
- Cc: bearheart@bearnet.com, brianp@apocalypse.org, kaimuse@apocalypse.org, mwm@contessa.phone.net, paulk@crocker.com, www-talk@w3.org
When I configured my system I specified a domain name, an IP address, a
set of protocols, etc. I don't see that the fact I could lie (if I so
chose) and use (for example) an incorrect domain name is wrong. The
situation with a web browser seems analogous. A user can be dishonest and
enter an invalid string, but that is the user's responsibility.
Switching gears a bit, I would agree that content negotiation will work
properly when I am able to configure my browser to request any version
available on the server. If I want frames (or tables, or push/pull, etc.)
then I should be able to specify as much via my configuration. The
problem is that [person configuring] the server is now in control of who
receives which version, and that is backwards.
---
Gregory Woodhouse gjw@best.com
I believe you are confusing two distinct things here.
Of course it is good to allow a power user, who knows what they
want, to specify via user-preference configuration fields
what capabilities they choose to take advantage of.
(Graphics vs textonly is one which already exists in some browsers.
Yes, there should be more options in all browsers.)
But that is not what the user-agent string is about at all.
It is for the non-power user, who wants things to work correctly
*automatically*. (Well-designed software is supposed to make things easy...)
The user-agent string is a way for telling a website what the
browser IS capable of, so that the pages displayed will be compatible
with the browser, with no effort from the user.
Completely different from custom configuration.
Brian
Received on Saturday, 27 January 1996 17:17:00 UTC