Opera and $$('.wrapper .class') seems not to work correctly
1 view
Skip to first unread message
Cyrus
Jan 12, 2009, 10:37:00 AM1/12/09
to Prototype & script.aculo.us
Hi there,
I have found a weird behaviour in the opera browser. I am not sure if
this is really a bug or an intended behaviour.
The following example shows that the $$ function only returns one
element instead of two. This is only because the second element is in
a different dom level. All other browser return two elements with this
example.
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/
TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" dir="ltr" xml:lang="de">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Test</title>
<script src="jsLib/prototype.js" type="text/javascript"></script>
<script type="text/javascript">
//<![CDATA[
function init() {
// init more information handling
var test = $$('.wrapper .class');
}
// when the dom is fully loaded, execute these scripts
document.observe("dom:loaded", init);
//]]>
</script>
</head>
<body>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
<div>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
</div>
</body>
</html>
I have found a weird behaviour in the opera browser. I am not sure if
this is really a bug or an intended behaviour.
The following example shows that the $$ function only returns one
element instead of two. This is only because the second element is in
a different dom level. All other browser return two elements with this
example.
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/
TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" dir="ltr" xml:lang="de">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Test</title>
<script src="jsLib/prototype.js" type="text/javascript"></script>
<script type="text/javascript">
//<![CDATA[
function init() {
// init more information handling
var test = $$('.wrapper .class');
}
// when the dom is fully loaded, execute these scripts
document.observe("dom:loaded", init);
//]]>
</script>
</head>
<body>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
<div>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
</div>
</body>
</html>
kangax
Jan 12, 2009, 11:30:48 AM1/12/09
to Prototype & script.aculo.us
On Jan 12, 10:37 am, Cyrus <arianglan...@googlemail.com> wrote:
> Hi there,
>
> I have found a weird behaviour in the opera browser. I am not sure if
> this is really a bug or an intended behaviour.
[...]
> Hi there,
>
> I have found a weird behaviour in the opera browser. I am not sure if
> this is really a bug or an intended behaviour.
Which versions of Prototype.js and Opera?
--
kangax
Cyrus
Jan 12, 2009, 12:27:14 PM1/12/09
to Prototype & script.aculo.us
Prototype 1.6.0.3
Opera 9.63
Opera 9.63
Alex Mcauley
Jan 13, 2009, 4:28:19 AM1/13/09
to prototype-s...@googlegroups.com
if its any help the same thing happens in google chrome !!
Richard Quadling
Jan 13, 2009, 4:41:22 AM1/13/09
to prototype-s...@googlegroups.com
2009/1/13 Alex Mcauley <webm...@thecarmarketplace.com>:
Adding ...
test.each(function(el){el.style.backgroundColor='green';});
to the script (just after the var test line) and I see 2 green divs.
Using Windows XP Pro SP3 with Chrome V1.0.154.43, Firefox V3.0.5 and
IE V7.0.5730.13.
NFF/Picnic?
--
-----
Richard Quadling
Zend Certified Engineer : http://zend.com/zce.php?c=ZEND002498&r=213474731
"Standing on the shoulders of some very clever giants!"
test.each(function(el){el.style.backgroundColor='green';});
to the script (just after the var test line) and I see 2 green divs.
Using Windows XP Pro SP3 with Chrome V1.0.154.43, Firefox V3.0.5 and
IE V7.0.5730.13.
NFF/Picnic?
--
-----
Richard Quadling
Zend Certified Engineer : http://zend.com/zce.php?c=ZEND002498&r=213474731
"Standing on the shoulders of some very clever giants!"
Cyrus
Jan 13, 2009, 4:52:28 AM1/13/09
to Prototype & script.aculo.us
Not in Opera 9.63 I am afraid ...
On 13 Jan., 10:41, "Richard Quadling" <rquadl...@googlemail.com>
wrote:
> 2009/1/13 Alex Mcauley <webmas...@thecarmarketplace.com>:
On 13 Jan., 10:41, "Richard Quadling" <rquadl...@googlemail.com>
wrote:
> 2009/1/13 Alex Mcauley <webmas...@thecarmarketplace.com>:
T.J. Crowder
Jan 13, 2009, 5:35:37 AM1/13/09
to Prototype & script.aculo.us
Hi,
> I am not sure if
> this is really a bug or an intended behaviour.
If you're getting back one element, not two, it's a bug. The question
is, in what -- Prototype's selection code, Opera, or both? I lean
toward Opera bug.
Interestingly, this _doesn't_ fail on Opera 9.10 for Windows: It
returns two elements as expected.
Have you tried it using a different classname than "class" for the
inner element? (I tend to think you just used that in your example,
having seen this in something else, but just in case...)
--
T.J. Crowder
tj / crowder software / com
Independent Software Engineer, consulting services available
> I am not sure if
> this is really a bug or an intended behaviour.
is, in what -- Prototype's selection code, Opera, or both? I lean
toward Opera bug.
Interestingly, this _doesn't_ fail on Opera 9.10 for Windows: It
returns two elements as expected.
Have you tried it using a different classname than "class" for the
inner element? (I tend to think you just used that in your example,
having seen this in something else, but just in case...)
--
T.J. Crowder
tj / crowder software / com
Independent Software Engineer, consulting services available
Cyrus
Jan 13, 2009, 6:26:19 AM1/13/09
to Prototype & script.aculo.us
I just used .class for this example. In my case I have a much more
complex HTML structure so it took me a while to actually find the bug.
> If you're getting back one element, not two, it's a bug. The question
> is, in what -- Prototype's selection code, Opera, or both? I lean
> toward Opera bug.
I guess the same, but maybe Prototype need an internal Opera
workarround here?
What would you suggest? File a bug report at Opera, Prototype or both?
complex HTML structure so it took me a while to actually find the bug.
> If you're getting back one element, not two, it's a bug. The question
> is, in what -- Prototype's selection code, Opera, or both? I lean
> toward Opera bug.
workarround here?
What would you suggest? File a bug report at Opera, Prototype or both?
kangax
Jan 13, 2009, 8:20:37 AM1/13/09
to Prototype & script.aculo.us
On Jan 13, 6:26 am, Cyrus <arianglan...@googlemail.com> wrote:
[...]
> What would you suggest? File a bug report at Opera, Prototype or both?
I can confirm this with 9.61 on Mac OSX. Here's a minimal failing test
case:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<title></title>
<script src="http://ajax.googleapis.com/ajax/libs/prototype/
1.6.0.3/prototype.js" type="text/javascript"></script>
document.write($$('.wrapper .class').length);
</script>
</body>
</html>
Should write "2" but writes "1".
This looks like an Opera bug. Moreover, I think we already have a
similar ticket filed (which was essentially due to a bug in Opera's
XPath engine bug and the way it handles descendant selectors). For
example, switching to "child" selector ('.wrapper > .class') returns
proper result - "2".
[...]
--
kangax
[...]
> What would you suggest? File a bug report at Opera, Prototype or both?
case:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<title></title>
<script src="http://ajax.googleapis.com/ajax/libs/prototype/
1.6.0.3/prototype.js" type="text/javascript"></script>
</head>
<body>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
<div>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
</div>
<script type="text/javascript">
<body>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
<div>
<div class="wrapper">
<a href="#" class="class">test</a>
</div>
</div>
document.write($$('.wrapper .class').length);
</script>
</body>
</html>
Should write "2" but writes "1".
This looks like an Opera bug. Moreover, I think we already have a
similar ticket filed (which was essentially due to a bug in Opera's
XPath engine bug and the way it handles descendant selectors). For
example, switching to "child" selector ('.wrapper > .class') returns
proper result - "2".
[...]
--
kangax
Cyrus
Jan 13, 2009, 8:23:51 AM1/13/09
to Prototype & script.aculo.us
Ok, thanks.
Unfortunately I cannot use the child selector in my case.
I'll write a bug report.
Unfortunately I cannot use the child selector in my case.
I'll write a bug report.
T.J. Crowder
Jan 13, 2009, 8:30:12 AM1/13/09
to Prototype & script.aculo.us
Hi Juriy,
That inline script could muddy the water in terms of DOM availability,
etc. Perhaps a minor mod:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<title></title>
<script src="http://ajax.googleapis.com/ajax/libs/prototype/
1.6.0.3/prototype.js" type="text/javascript"></script>
<script type="text/javascript">
document.observe('dom:loaded', function() {
$('matchcount').update('Matches: ' + $$
('.wrapper .inner').length);
});
</div>
</div>
<div id='matchcount'></div>
</body>
</html>
I also changed "class" to "inner", again for clear water.
FWIW,
-- T.J.
That inline script could muddy the water in terms of DOM availability,
etc. Perhaps a minor mod:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<title></title>
<script src="http://ajax.googleapis.com/ajax/libs/prototype/
1.6.0.3/prototype.js" type="text/javascript"></script>
document.observe('dom:loaded', function() {
$('matchcount').update('Matches: ' + $$
('.wrapper .inner').length);
});
</script>
</head>
<body>
<div class="wrapper">
<a href="#" class="inner">test</a>
</head>
<body>
<div class="wrapper">
</div>
<div>
<div class="wrapper">
<a href="#" class="inner">test</a>
<div>
<div class="wrapper">
</div>
</div>
<div id='matchcount'></div>
</body>
</html>
I also changed "class" to "inner", again for clear water.
FWIW,
-- T.J.
Cyrus
Jan 13, 2009, 8:50:40 AM1/13/09
to Prototype & script.aculo.us
kangax
Jan 13, 2009, 9:17:04 AM1/13/09
to Prototype & script.aculo.us
On Jan 13, 8:30 am, "T.J. Crowder" <t...@crowdersoftware.com> wrote:
> Hi Juriy,
>
> That inline script could muddy the water in terms of DOM availability,
> etc. Perhaps a minor mod:
I usually try to eliminate as much of irrelevant prototypeism as
> Hi Juriy,
>
> That inline script could muddy the water in terms of DOM availability,
> etc. Perhaps a minor mod:
possible when creating a testcase : ) Why use "dom:loaded" (which is
also known to be unreliable in IE in its current implementation) when
we can simply run script at the time when <body> is definitely loaded
and parsed.
[...]
--
kangax
T.J. Crowder
Jan 13, 2009, 12:58:41 PM1/13/09
to Prototype & script.aculo.us
> I usually try to eliminate as much of irrelevant prototypeism as
> possible when creating a testcase : )
Fair 'nuff, but I'd still use something other than inline to allow for
> possible when creating a testcase : )
browsers being difficult about the closing body tag and DOM rendering
time. Probably me being paranoid. Perhaps (shock and horror) DOM0
window.onload stuff, image loading not being an issue:
<script type='text/javascript'>
window.onload = function() {
alert('Found ' + $$('.wrapper .inner').length);
});
</script>
> at the time when <body> is definitely loaded
> and parsed.
about this stuff than I do.)
-- T.J. ;-)
kangax
Jan 13, 2009, 1:48:53 PM1/13/09
to Prototype & script.aculo.us
On Jan 13, 12:58 pm, "T.J. Crowder" <t...@crowdersoftware.com> wrote:
> > I usually try to eliminate as much of irrelevant prototypeism as
> > possible when creating a testcase : )
>
> Fair 'nuff, but I'd still use something other than inline to allow for
> browsers being difficult about the closing body tag and DOM rendering
> time. Probably me being paranoid. Perhaps (shock and horror) DOM0
> window.onload stuff, image loading not being an issue:
>
> <script type='text/javascript'>
> window.onload = function() {
> alert('Found ' + $$('.wrapper .inner').length);});
>
> </script>
This is perfectly fine of course : )
> > I usually try to eliminate as much of irrelevant prototypeism as
> > possible when creating a testcase : )
>
> Fair 'nuff, but I'd still use something other than inline to allow for
> browsers being difficult about the closing body tag and DOM rendering
> time. Probably me being paranoid. Perhaps (shock and horror) DOM0
> window.onload stuff, image loading not being an issue:
>
> <script type='text/javascript'>
> window.onload = function() {
> alert('Found ' + $$('.wrapper .inner').length);});
>
> </script>
>
> > at the time when <body> is definitely loaded
> > and parsed.
>
> Is it, definitely? Within the tag? (Serious question, you know more
> about this stuff than I do.)
most if not all other elements - exception being, maybe, <meta> tag
which has "Content-Type" and so might be parsed as early as
possible).
When a <script> element (the one before closing </body> tag) is
downloaded, parsed and evaluated, all of the DOM elements before it
should be present in DOM (that includes <body> of course)
I'm far from an expert in this field, but am pretty sure this is
what's happening : )
[...]
--
kangax
Jerod Venema
Jan 13, 2009, 4:43:17 PM1/13/09
to prototype-s...@googlegroups.com
Hey all, when you're replying to these threads, can you make sure you keep it on-list? I like to keep up with various solutions to all these problems, and seeing half the solution (or no solution at all) is really frustrating.
(@Kangax, I know I've seen a few replies to messages you've sent off-list, and since I know you've got good replies, I'd love it if they'd stay on the list. :-D)
Thanks!--
Jerod Venema
Frozen Mountain Software
http://www.frozenmountain.com/
919-368-5105
(@Kangax, I know I've seen a few replies to messages you've sent off-list, and since I know you've got good replies, I'd love it if they'd stay on the list. :-D)
Thanks!
Jerod Venema
Frozen Mountain Software
http://www.frozenmountain.com/
919-368-5105
Reply all
Reply to author
Forward
0 new messages