Bug 76931 – RenderSVGShape::strokeContains will fail for some strokes

• Status: RESOLVED
• Resolution: FIXED
• Component: SVG

[Stephen Chenney]

The method RenderSVGShape::strokeContains has a special case for zero-length-subpaths, and this only handles square end caps. It will give incorrect results for round endcaps on zero-length subpaths.

[Dirk Schulze]

Adding Rob who initially worked on zero-length-subpaths and line styles.

[Stephen Chenney]

Created attachment 125705 [details]
Patch

A simple fix for this issue and test cases.

[Darin Adler]

Comment on attachment 125705 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=125705&action=review

> LayoutTests/svg/hittest/zero-length-butt-cap-path-expected.txt:1
> +PASSED; PASSED; PASSED; PASSED; PASSED; PASSED; PASSED; PASSED;

Test output is a little cryptic. Tests that make it clear in the output what they are testing are even better.

> LayoutTests/svg/hittest/zero-length-butt-cap-path-expected.txt:6
>  Added: svn:eol-style
>    + LF

We normally use eol-style native instead of LF. Not sure why you’re doing it differently on these files.

[Stephen Chenney]

Comment on attachment 125705 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=125705&action=review

>> LayoutTests/svg/hittest/zero-length-butt-cap-path-expected.txt:1
>> +PASSED; PASSED; PASSED; PASSED; PASSED; PASSED; PASSED; PASSED;
> 
> Test output is a little cryptic. Tests that make it clear in the output what they are testing are even better.

I agree. I was uncertain about the balance of more JS code to construct good test output, vs test compactness. I'll play around with making things clearer, possibly converting all of the svg/hittest tests to use the JS test harnesses that other svg content uses.

>> LayoutTests/svg/hittest/zero-length-butt-cap-path-expected.txt:6
>>    + LF
> 
> We normally use eol-style native instead of LF. Not sure why you’re doing it differently on these files.

Probably just the result of editing in vim on linux. I didn't do anything special at all, or maybe it's a result of copying a problematic existing file as a starting point.

[WebKit Review Bot]

Comment on attachment 125705 [details]
Patch

Clearing flags on attachment: 125705

Committed r106882: <http://trac.webkit.org/changeset/106882>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.