272630 - On 'svg' elements getScreenCTM behaves differently to other implementations

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Jonathan Watt [:jwatt]]

Bug go make sure I follow up on this:
http://lists.w3.org/Archives/Public/www-svg/2004Sep/0066.html

Comment 1

[Jonathan Watt [:jwatt]]

Attachment: patch

Okay, this is how I think we should behave, which happens to be the way the
ASV6 alpha behaves (bar a bug I found in it). Certainly we should be accounting
for the transform due to the x and y attributes. I also think we should be
appending the viewBox to viewport transform. There is a note in the source
saying that if we do that for nsSVGSVGElement we need to modify
nsSVGGraphicsElement::GetCTM, but I don't see why. The only time
nsSVGGraphicElement::GetCTM effects nsSVGSVGElement::GetCTM or visa versa is in
nsSVGSVGElement::GetCTM.

Comment 2

[Jonathan Watt [:jwatt]]

Comment on attachment 170818 [details] [diff] [review]
patch

cancelling review request since we can't just use x and y (they are ignored on
outer <svg> elements)

Comment 3

[Jonathan Watt [:jwatt]]

I've followed up this issue in the thread beginning at
http://lists.w3.org/Archives/Public/www-svg/2005Jan/thread.html#15

Comment 4

[Jonathan Watt [:jwatt]]

CC'ing Chris and Dean. I haven't recieved any confirmation from the SVG WG that
this is an issue, or alternatively an explaination of why it isn't an issue. If
it is agreed that there is ambiguity here, can we have some clarification on
what the correct behaviour is please?

Comment 5

[Jonathan Watt [:jwatt]]

Attachment: testcase 1 - getScreenCTM on 'svg'

Comment 6

[Jonathan Watt [:jwatt]]

Attachment: testcase 2 - getScreenCTM on 'circle'

Comment 7

[Jonathan Watt [:jwatt]]

Attachment: testcase 3 - getScreenCTM on 'circle' with a 'translate' attribute

Comment 8

[Jonathan Watt [:jwatt]]

Attachment: testcase 4 - getScreenCTM on SVG embedded inline in XHTML

Comment 9

[Jonathan Watt [:jwatt]]

Attachment: unfinished snapshot

The changes to GetScreenCTM are "finished" and pass all the tests I just
attached. GetCTM and GetTransformToElement still need work.

Comment 10

[Jonathan Watt [:jwatt]]

Attachment: probably finished patch

This is as far as I'm going with this tonight. I'm fairly confident this is
correct, but I'll do some more testing tomorrow.

Comment 11

[Jonathan Watt [:jwatt]]

Attachment: testcase 5 - getScreenCTM on 'circle' with currentScale/Translate changed

Comment 12

[Jonathan Watt [:jwatt]]

Comment on attachment 197790 [details] [diff] [review]
probably finished patch

Requesting approval. This bug fixes a high visibility conformance issue. The
modified methods are not run by our internal code, they are only used by
content. The risk from this change should be very low.

Comment 13

[Asa Dotzler [:asa]]

Comment on attachment 197790 [details] [diff] [review]
probably finished patch

last day for non-critical changes.