Bug 56245 – Text on path positioning at zero startOffset

• Component: SVG
• Status: RESOLVED FIXED

[honyk]

Created attachment 85577 [details]
Test Case

When text (a single character) is positioned on the path with text-anchor set to middle and startOffset set to zero percent, its axis is not perpendicular to the path (the angle varies according to the radius) as can be seen in the attached test case.
As both Chrome and Safari are affected, I think it is a WebKit problem.
I would expect MSIE 9RC/Opera 11 behaviour.

[Rob Buis]

Created attachment 87035 [details]
Patch

[Nikolas Zimmermann]

Comment on attachment 87035 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=87035&action=review

Great catch, r=me with one change:

> Source/WebCore/platform/graphics/Path.cpp:108
> +    if (!length)
> +        traversalState.m_desiredLength = FLT_EPSILON;
> +    else
> +        traversalState.m_desiredLength = length;

I'd use here traversalState.m_desiredLength = length ? length : std::numeric_limits<float>::epsilon())

[WebKit Review Bot]

http://trac.webkit.org/changeset/82061 might have broken GTK Linux 32-bit Release
The following tests are not passing:
svg/W3C-SVG-1.1/animate-elem-08-t.svg

[Rob Buis]

Landed in r82061.

[Yuta Kitamura]

Since r82061, svg/W3C-SVG-1.1/animate-elem-08-t.svg started to fail (both on WebKit and Chromium's waterfall).

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20-%20chromium.org&tests=svg%2FW3C-SVG-1.1%2Fanimate-elem-08-t.svg&showExpectations=true

Do you know something?

[Rob Buis]

Hello Yuta,

(In reply to comment #5)
> Since r82061, svg/W3C-SVG-1.1/animate-elem-08-t.svg started to fail (both on WebKit and Chromium's waterfall).
> 
> http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20-%20chromium.org&tests=svg%2FW3C-SVG-1.1%2Fanimate-elem-08-t.svg&showExpectations=true
> 
> Do you know something?

You are right, sorry! But it turns out it is an improvement ; if you try without you'll see the triangle is wrongly aligned at the start, which is exactly the bug that got fixed (the text path aligning code uses the same Path subroutine as the animation code here). So if you want to update the results on chromium and mac please do. If you can't do the mac results yourself I can do that part.
Regards,

Rob.

[Yuta Kitamura]

(In reply to comment #6)
> You are right, sorry! But it turns out it is an improvement ; if you try without you'll see the triangle is wrongly aligned at the start, which is exactly the bug that got fixed (the text path aligning code uses the same Path subroutine as the animation code here). So if you want to update the results on chromium and mac please do. If you can't do the mac results yourself I can do that part.

Thanks! I will update test results for Chromium.

Unfortunately, I don't have working Mac environment now, so I cannot update the Mac results. Could you update them?

[Rob Buis]

Hello Yuta,

(In reply to comment #7)
> (In reply to comment #6)
> > You are right, sorry! But it turns out it is an improvement ; if you try without you'll see the triangle is wrongly aligned at the start, which is exactly the bug that got fixed (the text path aligning code uses the same Path subroutine as the animation code here). So if you want to update the results on chromium and mac please do. If you can't do the mac results yourself I can do that part.
> 
> Thanks! I will update test results for Chromium.
> 
> Unfortunately, I don't have working Mac environment now, so I cannot update the Mac results. Could you update them?

Sure, no problem.
Regards,

Rob.