445291 - SVG linearGradient not always rendered

Status: RESOLVED INVALID

(Core :: SVG, defect)

Description

[vezquex]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.0.1) Gecko/2008070208 Firefox/3.0.1
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.0.1) Gecko/2008070208 Firefox/3.0.1

A use element references a rect element in the defs tag. The rect element is to be filled by a linearGradient. The previous setup is generated by JavaScript in two slightly different ways and yet only the first way works. Both techniques work in Opera 9.5.

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Robert Longson [:longsonr]]

Something in your javascript code is converting linearGradient to lineargradient for myGradient2. SVG is case sensitive.

Comment 2

[Jonathan Watt [:jwatt]]

Thanks for your report vezquex.

I couldn't believe that Opera was case-insensitive so I dug into this a little. It seems like this comes down to Mozilla normalizing element names in HTML whereas Opera preserves the case. I'm sure there must be a bug open on this, but I couldn't find it.

More detail on what's happening in this particular testcase: some way under the svg.add(defs, ...) call, the 'clean' function in jquery-1.2.6.js (at the line "Go to html and back") appears to convert the 'linearGradient' element to an *HTML* node, then back to a string. I'd assume it's somewhere here (or somewhere similar) that "linearGradient" gets converted to "lineargradient".

CC'ing a few folks who might know more about that.

Comment 3

[Robert Longson [:longsonr]]

The testcase in the URL seems to work now. What change did you make vezquex?

Comment 4

[Boris Zbarsky [:bzbarsky]]

> It seems like this comes down to Mozilla normalizing element names in HTML
> whereas Opera preserves the case. I'm sure there must be a bug open on this,
> but I couldn't find it.

Last I checked, the normalizing behavior is correct (pending whatever HTML5 does with svg integration).

Comment 5

[vezquex]

Sorry, I was working around the problem by replacing 'lineargradient' with 'linearGradient'. The original test case has been restored now.