539576 - SVGTransform matrix changes still not live in all circumstances

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Robert Longson [:longsonr]]

Attachment: patch

setRotate, setSkewX, setSkewY create a new matrix so any references no longer update.

setting a value on the transform's matrix needs to turn the transform into a matrix transform.

Comment 1

[Robert Longson [:longsonr]]

With this patch (together with bug 512636) we pass the following tests:

http://dev.w3.org/SVG/profiles/1.1F2/test/svg/coords-dom-01-f.svg
http://dev.w3.org/SVG/profiles/1.1F2/test/svg/coords-dom-02-f.svg

Comment 2

[Robert Longson [:longsonr]]

Attachment: fix tests and prevent infinities

Comment 3

[Robert Longson [:longsonr]]

Jonathan, can you take up with the SVG WG what the correct thing to do is if you set a skewX or skewY of 90 or some other value that would cause us to store infinities in a matrix?

Comment 4

[Robert Longson [:longsonr]]

BTW I think this can land without answering comment 3 really, providing it gets an r+ of course. After all we test for infinities throughout the code and this just adds some missing tests and returns the same code.

Comment 5

[Jonathan Watt [:jwatt]]

Comment on attachment 421596 [details] [diff] [review]
fix tests and prevent infinities

r=jwatt

I think I'd rather "eps" and "epsilon" were called "tolerance", and perhaps add a short comment stating what it's for/why it's necessary for the uninitiated. Up to you though.

Comment 6

[Jonathan Watt [:jwatt]]

I also like the fact that the SVGMatrix object now doesn't change - it's more consistent with other parts of SVG (and more efficient in this case). :-)

Comment 7

[Jonathan Watt [:jwatt]]

I send a message to the SVG WG as requested in comment 3:

http://lists.w3.org/Archives/Public/www-svg/2010Jan/0085.html

Comment 8

[Robert Longson [:longsonr]]

(In reply to comment #5)
> I think I'd rather "eps" and "epsilon" were called "tolerance", and perhaps add
> a short comment stating what it's for/why it's necessary for the uninitiated.
> Up to you though.

Done.

pushed http://hg.mozilla.org/mozilla-central/rev/5f41d14731bc

Comment 9

[Robert Longson [:longsonr]]

reftest failures.

Comment 10

[Robert Longson [:longsonr]]

pushed http://hg.mozilla.org/mozilla-central/rev/7808d51d7696