Detecting browsers correctly

Martin Kopta is certainly right when he says that
 the document.all collection doesn’t uniquely identify
 browsers in the Internet Explorer family.

The problem arises when, on
 the client side, you need to find out the browser version and decide which piece of code to run, for example. It’s not uncommon for
 a browser to lack or incorrectly implement 
 standard DOM methods
 or other technologies, so you have to use its own browser-specific solution.

One of the most common ways to detect browsers in the Internet Explorer family
 is to test for the document.all collection. This usually takes 
 the following form:

if(!document.all){
     //standardni vetev kódu
   }else{
     //proprietární využití pro IE
   }

But that very construction sends browsers such as Opera down
 the Internet Explorer branch. When it sees document.all
, Opera returns an Object in response, so it can get into the part of the code that
 may cause an error.

The lesson is that it pays to test for features that really are specific, or proprietary, to
 the particular JavaScript or JScript implementation or technology. 
 You should be able to detect Opera reliably by checking window.opera. Internet
 Explorer 4.01 can be detected using window.print. For later versions
 of Internet Explorer I’d recommend its particular implementation of events instead 
 (the Event object). For more on the subject, see
 Eddie Traversa’s article 
 
 Object Detection in the New Browser Age

The question remains how to detect newer versions that may drop their proprietary 
 technologies. For reasons such as backward compatibility, however, we can’t
 expect that to happen. I’d tackle the problem case by case. There’s always
 a bug or difference to be found …