I'm not 100% sure, but I think you misread Matt's post. Matt seemed to be saying the same thing you are--the order of keys should not be relied upon.
On Sep 2, 4:26 pm, "Michael Geary" <[EMAIL PROTECTED]> wrote: > > > Matt Kruse wrote: > > > > This appears to be a bad assumption in the jQuery tests. > > > > The code in param() calls: > > > for ( var j in a ) > > > and makes the assumption that the keys will be returned in the same > > > order they are specified. This is not an assumption that should be > > > made, so the test should be changed to allow for the > > > returned string to be in arbitrary key order. > > From: Guy Fraser > > > I've never seen an ECMA script compiler (JS, AS, etc) that > > doesn't iterate through named references in the expected order? > > But there is no "expected order" in a for..in loop. Any decent JavaScript > reference book, such as Flanagan's, should point this out. > > Of course, you could make the same argument that I made for the text node > issue, that too much code depends on this and they have to fix it. But I > think given their architecture for property lookups, it may not be possible. > And I would hope that not too much code does depend on it anyway. > > -Mike