Daniel Schierbeck wrote:

There are two problems with your 'hash' proposal. First of all it is very javascript oriented. Your 'hash' is really just an object which you access properties on, there is no way that would work in for example C++.

The other problem is that using js-objects as hashes has issues. For example your hash would come pre-populated with some entries like 'toString', 'toSource' and 'watch'. You would also be unable to map the prefixes '__parent__' and '__count__' to anything since those properties are ReadOnly.

While these things aren't usually a problem since you're unlikly to use those prefixes, it's not something I would want to bake into the standard.

So I suggest you just use a hash inside your XPathNSResolver function. If you write clean code, rather then the one-liner above, it'll look much simpler.

function resolver(str) { hash = {xh: 'http://www.w3.org/xhtml', xl: 'http://www.w3.org/1999/xlink'} return hash[str]; } document.selectorsSelectElements('xh|p', resolver);

/ Jonas