Bug 83568 – [SVG] Nothing should be stroked when the stroke-width is 0

• Component: New Bugs
• Status: RESOLVED FIXED

[Pierre Rossi]

[SVG] stroke-width:0 should be treated like "none"

[Pierre Rossi]

Unless I'm overlooking something, it seems to me when I read the spec [1] that we shouldn't leave it up to each individual port to ensure they respect it. In this particular case, I fail to see how delegating this decision does any good: it induces overhead at best, and non-compliance if we're a little less lucky.
In the case of the Qt port this was an issue: QPainter treats a zero-width pen as a cosmetic pen, which still caused that bug to be present.

[1] http://www.w3.org/TR/SVG/painting.html#StrokeWidthProperty

[Pierre Rossi]

Created attachment 136440 [details]
Patch

[Alexis Menard (darktears)]

(In reply to comment #1)
> Unless I'm overlooking something, it seems to me when I read the spec [1] that we shouldn't leave it up to each individual port to ensure they respect it. In this particular case, I fail to see how delegating this decision does any good: it induces overhead at best, and non-compliance if we're a little less lucky.
> In the case of the Qt port this was an issue: QPainter treats a zero-width pen as a cosmetic pen, which still caused that bug to be present.
> 
> [1] http://www.w3.org/TR/SVG/painting.html#StrokeWidthProperty

Well I disagree a bit the spec says : "A zero value causes no stroke to be painted." but that doesn't mean it doesn't have a stroke, it has a stroke property defined but the stroke width is 0 and in that case the spec define how it should renders.

[Pierre Rossi]

@Alexis: ah, you're absolutely right. Ok, let's try something else then :)

[Pierre Rossi]

Created attachment 136472 [details]
Patch

[WebKit Review Bot]

Comment on attachment 136472 [details]
Patch

Attachment 136472 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/12386048

New failing tests:
svg/custom/js-update-style.svg
svg/custom/path-zero-strokewidth.svg

[WebKit Review Bot]

Created attachment 136570 [details]
Archive of layout-test-results from ec2-cr-linux-04

The attached test failures were seen while running run-webkit-tests on the chromium-ews.
Bot: ec2-cr-linux-04  Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'>  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick

[Pierre Rossi]

It seems Chromium has the same issue.

[Nikolas Zimmermann]

Comment on attachment 136472 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=136472&action=review

Thanks for tackling this, some comments leading to r-:

> Source/WebCore/rendering/svg/RenderSVGRect.cpp:123
> +    if (style()->svgStyle()->strokeWidth().isZero())
> +        return;

How about adding "hasVisibleStroke" to SVGRenderStyle, which returns "hasStroke() && !strokeWidth().isZero()".
Then this could turn into: if (!style()->svgStyle()->hasVisibleStroke()) return;

> Source/WebCore/rendering/svg/RenderSVGShape.cpp:104
> +    if (!style()->svgStyle()->strokeWidth().isZero())

Ditto.

> Source/WebCore/rendering/svg/RenderSVGShape.cpp:276
> +    if (this->style()->svgStyle()->strokeWidth().isZero())

Ditto, not listing further occurrences of this.

> LayoutTests/ChangeLog:7
> +

You should also mark this as needs rebaseline for chromium.
Look for the SVG Tests section in platform/chromium/test_expectations.txt, and add BUGWK83568 : svg/custom/path-zero-strokewidth.svg = IMAGE, same for the other failure.

[Pierre Rossi]

Created attachment 138791 [details]
Patch

[Nikolas Zimmermann]

Comment on attachment 138791 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=138791&action=review

r=me, but please make sure this builds after you landed it, as there are no EWS results available.

> Source/WebCore/rendering/svg/RenderSVGShape.cpp:276
> +    if (!this->style()->svgStyle()->hasVisibleStroke())

Can't you used the passed in RenderStyle* here? I think you can.

[Pierre Rossi]

(In reply to comment #11)

Landed manually with suggested change in r115201.

> (From update of attachment 138791 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=138791&action=review
> 
> r=me, but please make sure this builds after you landed it, as there are no EWS results available.

Alright, watching the bots now.
> 
> > Source/WebCore/rendering/svg/RenderSVGShape.cpp:276
> > +    if (!this->style()->svgStyle()->hasVisibleStroke())
> 
> Can't you used the passed in RenderStyle* here? I think you can.

And I think you're right :)

Thanks

[Stephen Chenney]

Committed r116992: <http://trac.webkit.org/changeset/116992>

[Stephen Chenney]

Updated Chromium expectations. Now we're really done.