Browser sniffing for customisation

'browser sniffing' - the use of HTTP information to determine what User
Agent is accessing a resource, and customising it accordingly, is already
common in some areas such as language negotiation. It has also been used,
and more often suggested, as a technique for tailoring websites to the
needs of specific users, which could provide benefits for accessibility.

My personal view is that there are big trade-offs which make it a poor
solution in most cases.

Determining which user agent is accessing the data is insufficient
information. For example, one University produced a Java-based navigation
system which it only supplied to netscape 3+ and explorer 4+, with other
User Agents getting an HTML-based system. This meant that users of lynx,
netscape 1.1 and opera all got a working system. But users of screen
readers were served something that they could not use, because of a flaw
in the reasoning. If explorer continues to support assistive technology,
(as it is reasonable to assume they will) then users of explorer cannot be
relied on to have the full multimedia capacity that the browser
<em>can</EM> provide. Which defeats the purpose of identify users with
that capacity.

Another, but far less compelling reason in theory, is maintenance. The use
of a text-only shadow of a website as a primary strategy requires more
maintenance in most cases, and in practise generally fails at some level,
where links go to a mutimedia version which is not accessible, or the
text-only version is not maintained as often as the 'main' version and
eventually becomes out of date.

I will however add this to the issues list.

charles


--Charles McCathieNevile -  mailto:charles@w3.org
phone: * +1 (617) 258 0992 *  http://purl.oclc.org/net/charles
       **** new phone number ***
W3C Web Accessibility Initiative -  http://www.w3.org/WAI
545 Technology sq., Cambridge MA, USA

Received on Tuesday, 29 December 1998 14:20:37 UTC