On Nov 23, 2011, at 17:45 , Jonas Sicking wrote: > On Wednesday, November 23, 2011, Robin Berjon <[email protected]> wrote: > > I would be thinking about something along the lines of (untested, off the > > top of my head): > > > > Node.prototype.queryXPath = function (xpath) { > > var snap = this.ownerDocument > > .evaluate(xpath, > > this, > > function () { return > > "http://www.w3.org/1999/xhtml";; }, > > XPathResult.ORDERED_NODE_SNAPSHOT_TYPE, > > null); > > var ret = []; > > for (var i = 0; i < snap.snapshotLength; i++) > > ret.push(snap.snapshotItem(i)); > > return ret; > > }; > > > > Which hardly strikes me as a terribly complex addition. > > I like this approach. Though I would make the type ANY and return whatever > type the expression produced.
I agree that in general ANY_TYPE is desirable so that we can get non-node results, the problem is that for node sets it defaults (for reasons that defy any understanding that I have) to UNORDERED_NODE_ITERATOR_TYPE — when we clearly want something ordered and a snapshot. But specifying around that is not a big deal. So, now for the money question: should we charter this? -- Robin Berjon - http://berjon.com/ - @robinberjon
