[thelist] [evolt] Is server-side browser-sniffing a bad idea? (was RE: js include error in IE 3)

aardvark roselli at earthlink.net
Thu Dec 6 08:25:29 CST 2001


> From: "George Dillon  <> Evolt!" <evolt at georgedillon.com>
[...]
> I've read recently that using server-side browser-sniffing to keep
> code lean by tailoring it for each browser's quirks is NOT a good idea
> because of web caches.  The undesirable scenario is that a web cache
> stores the page designed for WinMSIE5 (e.g.) and then delivers THAT
> verison when requested by a user with MacNN4... resulting in a layout
> mess for that particular user.
> 
> Is this a real issue or have I misunderstood how web caches work?
> 
> [Note: I am not talking about client browser caches but the server
> caches created by some ISPs.]

this is absolutely a concern... think of proxies at large 
organizations... they often cache pages, and depending on who 
requested it, that's what you see...

a previous employer of mine didn't believe this and insisted on all 
sorts of browser-sniffing code to deliver pages with JS and HTML 
specific to the browser... a quick demonstration of me calling up 
the page in Navigator showed how all the IE users (which was 
nearly everyone) saw a page with some strange layout and lots of 
JS errors...

that it was an ecommerce site made the impact of the argument all 
that much more potent... suddenly, IE users in our organization 
couldn't place orders...

keep in mind, proxies and other caching thingies (ISPs, for 
example) can be configured to help alleviate this issue, or make it 
worse.... if you aren't configuring it, assume the worst...





More information about the thelist mailing list