Bah... you already said so... I should had read it all before
posting :P

--
Ariel Flesler
http://flesler.blogspot.com/

On 7 jul, 20:43, "Michael Geary" <[EMAIL PROTECTED]> wrote:
> The array returned by .get() does have its elements in the correct order.
> The bug is actually in your code:
>
> >         for(ai in arr) { txt.push(arr[ai].innerHTML); }
>
> You can't use a "for...in" loop on an array and expect to get consistent
> results. The order of enumeration is implementation dependent. It may be in
> numerical order, or it may be in the order the elements were added to the
> array, or anything else. You may note that you get different results in IE7
> and Firefox 3.
>
> A for...in loop also enumerates non-array-element properties of the array.
> For example, if any JavaScript code or library adds methods to
> Array.prototype (and many libraries do this), those methods will also be
> enumerated in your for...in loop.
>
> A numeric for loop would work correctly:
>
>     for( i = 0, n = arr.length;  i < n;  ++i )
>         txt.push(arr[i].innerHTML);
>
> The reason .each works correctly is that it does a numeric for loop like
> that.
>
> What changed in 1.2.6?
>
> The code you found in makeArray does assign the elements into the correct
> array indices but it does them backwards chronologically. IOW, makeArray is
> assigning elements in this order:
>
>     ret[2] = array[2];
>     ret[1] = array[1];
>     ret[0] = array[0];
>
> That's perfectly valid code, of course, but a for...in loop on the "ret"
> array will give inconsistent results. A correct numeric for loop will not be
> bothered by that.
>
> In 1.2.3 the loop in makeArray ran in forward order:
>
>     ret[0] = array[0];
>     ret[1] = array[1];
>     ret[2] = array[2];
>
> All that said, I wonder if it wise for makeArray to be running this loop in
> reverse, just because of problems like this. Would be that much less
> efficient to run the loop in forward order?
>
> There's some further discussion in the ticket for this code change:
>
> http://dev.jquery.com/ticket/2619
>
> -Mike
>
>
>
> > From: joelarson
>
> > I have just noticed that the order of elements returned by
> > .get() seem to have been reversed in 1.2.5 and 1.2.6.  
> > Interestingly, each seems to run in the expected order (from
> > first element to last as encountered in the document).  To
> > demonstrate:
>
> > ----------------------------------
> > <script src="http://jqueryjs.googlecode.com/files/jquery-1.2.5.js";></
> > script>
> > <span>1</span><span>2</span><span>3</span>
> > <script language=JavaScript><!--//
> >     $(function() {
> >         var arr=$("span").get();
> >         var txt=[];
> >         for(ai in arr) { txt.push(arr[ai].innerHTML); }
> >         alert("via get(): "+txt.join(","));
> >         $("span").each(function(i,o) { alert("via each #"+i
> > +".html="+o.innerHTML); });
> >     });
> > //--></script>
> > ----------------------------------
>
> > Your first alert will show "via get(): 3,2,1".  Your next 3
> > alerts will show 1,2,3 in order as the innerhtml.
>
> > This seems to come from around line 1133 of the
> > jquery-1.2.5.js (around the same line in 1.2.6), where the
> > way the array is filled in makeArray is:
>
> > ----------------------
> > while( i )
> >   ret[--i] = array[i];
> > ----------------------
> > (which obviously builds a backwards return array)
>
> > whereas in jquery-1.2.4.js line 1125 is code:
>
> > ----------------------
> > for ( var i = 0, length = array.length; i < length; i++ )
> >   ret.push( array[ i ] );
> > -----------------------
>
> > It seems to me that the get() must return the elements in the
> > document specified order.  Even though this is not spelled
> > out in the specification it seems to be the only rational behavior.
>
> > If my assumption is correct then, the easiest way to fix is to insert:
>
> > ------------
> > ret.reverse();
> > ------------
>
> > at line 1141 of 1.2.6.js (or equivalent in future 1.2.7).
>
> > I have not found any posts on this or notes anywhere, but it
> > is possible I do not know every relavent area to check.  So
> > don't beat me up if this is known or desired behavior-- just
> > tell me where I should have looked.  Thanks!- Ocultar texto de la cita -
>
> - Mostrar texto de la cita -

Reply via email to