Bug 76931

Summary: RenderSVGShape::strokeContains will fail for some strokes
Product: WebKit Reporter: Stephen Chenney <schenney>
Component: SVGAssignee: Stephen Chenney <schenney>
Status: RESOLVED FIXED    
Severity: Normal CC: krit, rwlbuis, schenney, webkit.review.bot, zimmermann
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on: 71820, 77964    
Bug Blocks:    
Attachments:
Description Flags
Patch none

Stephen Chenney
Reported 2012-01-24 12:36:32 PST
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.
Attachments
Patch (14.54 KB, patch)
2012-02-06 14:05 PST, Stephen Chenney 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Dirk Schulze
Comment 1 2012-01-24 14:03:51 PST
Adding Rob who initially worked on zero-length-subpaths and line styles.
Stephen Chenney
Comment 2 2012-02-06 14:05:26 PST
Created attachment 125705 [details] Patch A simple fix for this issue and test cases.
Darin Adler
Comment 3 2012-02-06 15:56:06 PST
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 4 2012-02-06 16:10:44 PST
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 5 2012-02-06 17:31:06 PST
Comment on attachment 125705 [details] Patch Clearing flags on attachment: 125705 Committed r106882: <http://trac.webkit.org/changeset/106882>
WebKit Review Bot
Comment 6 2012-02-06 17:31:16 PST
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.