890384 - Firefox adds a white edging inside this svg, does not occur in Opera/Chrome/IE (IE has another rendering bug)

Status: RESOLVED DUPLICATE of bug 1203376

(Core :: SVG, defect)

Description

[nemo]

I encountered this SVG w/ someone who was opinionating loudly about Firefox inadequacies with this SVG he had created as a key exhibit.  Apparently the light source was offset incorrectly in Firefox.

I examined it, and decided the offset was correct in Firefox, and was due to a translate of the light source.  I fixed this in the javascript by adding an offset to the x/y, which is the version you see above.

In attempting to prove my case that Chrome was in error here, I loaded the image in other browsers, and noticed a white border surrounding the SVG content in Firefox only (IE had a separate rendering error w/ bottom of image).

Now, on some Firefox versions, like the Firefox 22 stable + intel gfx card + linux on this machine here, the white border crawls in and out from edge of image as I move the light source around.  In others it seems fixed in place.

This SVG is truly horrible, a ton of needless translations.  Appears to be a side effect from an editing tool he used to extract the image from some larger one.  A kind of SVG stress test.

Oh. Also, the 3 balloon cutie mark embossing renders not-embossed or maybe just barely visibly embossed in Firefox.  I'm not sure that is a bug.

As an aside, I'm going to be sad when Presto dies, since loading things in Opera is most effective way to rebut a "works in Chrome therefore Firefox is broken" argument.  "Works in IE" just isn't as effective with people due to past history...

At least Webkit is forked into two incompatible rendering engines now.  So I suppose Safari is the fallback for tests like this.

Comment 1

[nemo]

Attachment: Demo svg

Comment 2

[Robert Longson [:longsonr]]

We really need a minimal testcase. This is too big to debug.

Comment 3

[nemo]

Aight.  I'll do the tried and true, delete chunks of the file, refresh thing, at some point.

Oh, I think maybe the cutie mark-possibly-not-being a bug was due to testing on various browsers.
Really don't remember.  So maybe you'd want to glance at that too.
When I get back to my machine w/ all my VMs I'll load it again.

Comment 4

[Robert Longson [:longsonr]]

Maybe you could spin the cutie mark off into a separate bug and create a minimised testcase for that too.

Comment 5

[Ciprian Muresan [:cmuresan], Ecosystem QA]

Hi,

Verified this issue on latest Firefox Release and could not reproduce it anymore. However, when testing this on Firefox v20.0 there was a difference in display that seems to be the one reported by you.
Performed regression testing and found out a more recent issue that looks like it has treated the same problem (bug 1203376). 
Considering this, I am marking this one as resolved-duplicate. Please feel free to reopen it if the issue is still present and provide more information.

Thanks,
Cipri