479058 - getBoundingClientRect returns bogus values for SVG elements

Status: RESOLVED DUPLICATE of bug 530985

(Core :: SVG, defect)

Description

[Marcus Better]

User-Agent:       Mozilla/5.0 (compatible; Konqueror/4.2; Linux) KHTML/4.2.0 (like Gecko)
Build Identifier: Mozilla/5.0 (X11; U; Linux x86_64; sv-SE; rv:1.9.0.6) Gecko/2009020407 Iceweasel/3.0.6 (Debian-3.0.6-1)

When calling getBoundingClientRect on an SVG element, the resulting box object indicates weird values. Usually its width and height are zero, even though the SVG element has positive width and height attributes.

In a more complex case I have also noticed that the returned coordinates may change unpredictably after the DOM is modified by adding some children to the SVG element (without changing the position or size of the SVG element itself). This happens both in Firefox 3.0 and Firefox 3.1 beta 2, but in the latter case the variation was much greater.


Reproducible: Always

Comment 1

[Marcus Better]

Attachment: test case

In FF 3.0 and 3.1b2, this produces:
  left=8, right=8, top=8, bottom=8

Comment 2

[Robert Longson [:longsonr]]

Please try a current trunk build.

Comment 3

[Marcus Better]

It may be noted that document.getBoxObjectFor(), in contrast, appears to return sensible values for SVG elements.

Comment 4

[Marcus Better]

>   left=8, right=8, top=8, bottom=8

Same result on Minefield: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2a1pre) Gecko/20090218 Minefield/3.2a1pre

Comment 5

[Robert Longson [:longsonr]]

getBoundingClientRect gets the area covered by SVG graphic shapes e.g. rect, circle, text etc. There are none here so the bounding box is and should be empty.

Comment 6

[Jonathan Watt [:jwatt]]

That wasn't how I thought it really worked, so I looked at http://www.w3.org/TR/cssom-view/#the-getclientrects

I think the relevant sentence is "Return a (static) TextRectangleList object containing a list of (static) TextRectangle objects in content order describing the border boxes, *even if they are empty*, for the element on which it was invoked."

Emphasis mine. So the border box of the <svg> element happens to coincide with the edges of the viewport established by the <svg> in this case since it has no 'border' of 'padding' here. So it would seem to me that it should be returning a non-empty rect, no?

Comment 7

[Robert Longson [:longsonr]]

What do you think roc?

Comment 8

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

You should refer to this version:
http://dev.w3.org/csswg/cssom-view/

See step 2 of getClientRects:
http://dev.w3.org/csswg/cssom-view/#the-getclientrects-and-getboundingclient

Comment 9

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

I'm not actually sure what the SVG bbox is in these testcases, off the top of my head. If our code disagrees with the cssom-view draft we should fix it, otherwise if you think the current draft is not what authors want, suggest a better way and we'll fix the draft and our code.

Comment 10

[Jonathan Watt [:jwatt]]

Okay. I don't have the time to look at this atm, so I'll reopen this to make sure I don't forget about it.