186516 - offsetWidth,clientWidth and scrollWidth properties return incorrect values with empty block elements

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[tfriesen]

These IE compatible properties seem to return incorrect values when there is no
content in a block element.

In the following testcase,the document.body contains a empty DIV element named
"div1",no styling is applied to the document.

offsetWidth,clientWidth and scrollWidth all return zero for div1,
but using getComputedStyle() for the width property gives a value.
These properties should be returning the same width values as the document.body
since the width values are "auto".
clientWidth also returns zero in IE6 for some reason.

offWidth also returns zero for the document.body.
clientWidth and scrollWidth both give a value.
These values are both greater then the computed width of the document.body,so 
that does'nt seem right.

Is the computed width of document.body suppose to be less than the width of the
root element?

The testcase should work in IE as well

Comment 1

[tfriesen]

Attachment: testcase:An alert onload will show property values

Comment 2

[tfriesen]

Attachment: testcase:same as above but with content added to "div1"

Comment 3

[Johnny Stenback (:jst)]

Mass-reassigning bugs to [e-mail]

Comment 4

[Mats Palmgren (inactive)]

*** Bug 198999 has been marked as a duplicate of this bug. ***

Comment 5

[Boris Zbarsky [:bzbarsky]]

The problem here is that the offset rect is computed by starting with an empty
rect and unioning it with every rect of the element (continuations, etc).

But in this case the element rect is 0 height, so empty.  Thus we fall into the
"both rects empty case" in UnionRect().

I'm guessing the best fix here is to copy the relevant code out of UnionRect and
do the union by hand.... roc, any better ideas?

Comment 6

[Robert O'Callahan (:roc) (email my personal email if necessary)]

nope, that sounds correct.

Comment 7

[Boris Zbarsky [:bzbarsky]]

Attachment: Patch

Comment 8

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Actually, I wonder if the right thing to do here is to special-case the empty
rect and if it is empty, set it to (first-in-flow.x, first-in-flow.y, 0, 0).

Comment 9

[Boris Zbarsky [:bzbarsky]]

You mean if the result is empty once we're done with the loop?

Comment 10

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Yeah. So if during the loop we find a non-empty rect, we use that.

Comment 11

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 149701 [details] [diff] [review]
Patch

minusing while we wait for bz to come back and respond

Comment 12

[Boris Zbarsky [:bzbarsky]]

Attachment: Per review comments

Comment 13

[Boris Zbarsky [:bzbarsky]]

Checked in.