jQuery Forum

Seems very good... are you going to do some assert-like tests? to
verify the calculations went well..
I like it.
Ariel Flesler

I might be misunderstanding your question but the actual test/offset.html does use assert-like tests, specifically equal, to insure that offset is actually returning the correct value. The visual test is just an extra layer to debug/test.
<div><br class="webkit-block-placeholder"></div><div>--</div><div>Brandon Aaron

Uhm.. weird, it said 0 tests/0ms, then refreshed, and they ran, then
refreshed again, and remained still.. whatever, the test is perfect :)
Ariel Flesler

Yeah there seems to be some timing issues caused by having to pop open a new window. I haven't been able to really track it down yet. It should run all the tests but not summarize properly ... at least in my experience so far.
--
Brandon Aaron

Can you add a test case for elements inside tables, especially tables
with a tbody that scrolls e.g., $("table tbody").css({height:"200px",
width: "200px", overflow: "auto"})
I am finding the offset is wrong for any child elements within a table
tbody that is scrolled. It's correct for any other elements that
scroll except for tbody.

I've been looking at this issue in firefox and it looks like the bug
was introduced when fixing a bug in opera. this patch addresses the
issue at least in firefox for jquery version 1.2.2 (not sure what the
opera impact is - haven't tested that).
Index: jquery.js
===================================================================
--- jquery.js    (revision 36998)
+++ jquery.js    (working copy)
@@ -3341,7 +3341,7 @@
            // Get parent scroll offsets
            while ( parent && parent.tagName && !/^body|html$/
i.test(parent.tagName) ) {
                // Remove parent scroll UNLESS that parent is inline or a table
to work around Opera inline/table scrollLeft/Top bug
-                if ( !/^inline|table.*$/i.test(jQuery.css(parent, "display")) )
+                if ( !/^inline|table.*[^group]$/i.test(jQuery.css(parent,
"display")) )
                    // Subtract parent scroll offsets
                    add( -parent.scrollLeft, -parent.scrollTop );
basically, when the tbody has been scrolled, you need to deduct that
scrollTop and scrollLeft. so for displays matching table-.+-group
(table-row-group, table-header-group...) those scrollTop/Lefts should
be respected.

nice! works for me!
btw I created a ticket for this tbody offset issue: http://dev.jquery.com/ticket/2268

Thanks for making the ticket. I'll get to it as soon as I can.
--
Brandon Aaron

On Dec 4 2007, 1:44 pm, "Brandon Aaron" <brandon.aa...@gmail.com>

Yeah all the browsers get offsetTop and offsetLeft wrong in one way or another. And it is even different between Safari 2 and Safari 3. The older offset method had a relative option that would return the offset relative to a given parent. The offset test suite is located in the jQuery SVN repository: <a href="http://dev.jquery.com/browser/trunk/jquery/test">http://dev.jquery.com/browser/trunk/jquery/test</a>
Also here is the previous version of the offset method: <a href="http://dev.jquery.com/browser/trunk/plugins/dimensions/jquery.dimensions.js?rev=2760#L379">http://dev.jquery.com/browser/trunk/plugins/dimensions/jquery.dimensions.js?rev=2760#L379</a>
BTW ... I like the idea of fixing offsetTop and offsetLeft ... please keep me in the loop on your progress. :)
--
Brandon Aaron