Result if far too different

[launchpad-issue-importer]

Optimizing loses some elements.

---

Imported from Launchpad using lp2gh.

• date created: 2011-05-19T16:04:38Z
• owner: andresm-plast
• the launchpad url was https://bugs.launchpad.net/bugs/785270

[launchpad-issue-importer]

(by andresm-plast)

[launchpad-issue-importer]

(by jazzynico)
Confirmed on Ubuntu 10.10, Inkcape trunk revision 10220 (scour 0.26).

• Two of the three boxes' gradient are converted to flat color.
• The bottom left connector's marker is badly placed on the right of the left box.

[launchpad-issue-importer]

(by jazzynico)

[launchpad-issue-importer]

(by louis-simard-deactivatedaccount)
Rendered in Firefox 3.6.17/Linux, I do see that 2 of the 3 rounded rectangles have lost their white gradient, but I don't see a bad marker. Where is it?

[launchpad-issue-importer]

(by jazzynico)
The marker issue may be due to a connectors bug in Inkscape.
Attached, a screenshot taken on Windows XP with Inkscape r10221.

[Ede123]

Marker issue is definitely fixed (I made sure of that in 842123a).

The gradient is still removed. I reduced the testcase and tracked down the issue:

<?xml version="1.0" encoding="UTF-8"?>
<svg height="600" width="600" xmlns="http://www.w3.org/2000/svg">
    <defs>
        <radialGradient id="a" cx="10%" cy="10%" r="100%" fx="10%" fy="10%">
            <stop offset="0%" stop-color="#ffffff" />
            <stop offset="100%" stop-color="#ffffcc" />
        </radialGradient>
        <radialGradient id="b" cx="10%" cy="10%" r="100%" fx="10%" fy="10%">
            <stop offset="0%" stop-color="#ffffff" />
            <stop offset="100%" stop-color="#ffffcc" />
        </radialGradient>
    </defs>
    <rect x="0" y="0" width="100" height="80" rx="10" ry="10" fill="url(#a) #ffffcc" />
    <rect x="0" y="0" width="100" height="80" rx="10" ry="10" fill="url(#b) #ffffcc" />
</svg>

The problem is the specification of a fallback background color in fill="url(#b) #ffffcc".
When scoured the identical gradients #a and #b are merged, but the reference is not updated properly. (As I was sceptical myself: This type of specification seems to be valid, see SpecifyingPaint, specifically the entry <funciri> [ none | currentColor | <color> [<icccolor>] ]