Linear Gradient produces "seam" line when shape to fill uses curves and decimal coords
Reported by
r...@chartiq.com,
Oct 17 2017
|
|||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_5) AppleWebKit/601.7.8 (KHTML, like Gecko) Version/9.1.3 Safari/537.86.7 Steps to reproduce the problem: Load the attached file and observe the result, the gradient does not fill correctly as it does in other browsers. What is the expected behavior? Gradient should fill smoothly without seam lines. What went wrong? Not sure, but see screenshots comparing Chrome (1.11.38 PM) vs Safari (1.12.53 PM). Firefox is similar to Safari. Did this work before? Yes It might have been 57. Does this work in other browsers? Yes Chrome version: 61.0.3163.100 (Official Build) (64-bit) Channel: stable OS Version: OS X 10.9.5 Flash Version: Shockwave Flash 25.0 r0
,
Oct 18 2017
Able to reproduce this issue on reported version 61.0.3163.100 and on latest canary 64.0.3242.0 using Mac 10.12.6, Window10 and Ubuntu 14.04 with steps mentioned in comment#0. Manual Bisect Info: =============== Good Build: 58.0.3006.0 Bad Build: 58.0.3007.0 You are probably looking for a change made after 449080 (known good), but no later than 449081 (first known bad). CHANGELOG URL: The script might not always return single CL as suspect as some perf builds might get missing due to failure. https://chromium.googlesource.com/chromium/src/+log/e14bcf85b26d7e95338f78ebc6ce72fedd5cb913..4fa27c5c19dadb9b737cb004c8ad176e8cb45422 Review-Url: https://codereview.chromium.org/2680003002 Suspecting same from changelog. senorblanco: Please confirm the bug and help in re-assigning if it is not related to your change. Thanks!
,
Nov 8 2017
Notes: - increasing the tolerance (to tolerance^3) in path_to_contours() causes it to go away - replacing the first quadraticCurveTo(400,1600,475,1453.5) with a lineTo(475,1453.5) fixes it - (equivalently, moving the tangent point to the endpoint fixes it: quadraticCurveTo(475,1453.5,475,1453.5)) So it seems to be related to the endpoint for the first quadratic segment.
,
Nov 8 2017
Here's a reduction which uses only linear segments and removes the linear gradient, but still shows the issue. I suspect that this is related on the intruding vertex problem, and the computed alpha near the intersection vertex is wrong.
,
Nov 8 2017
Note to self: this is fixed by the straight skeleton phase 3 (WIP) when outer contour overlap reduction is turned on.
,
Nov 13 2017
Since similar regressions are Pri-2 (and the fix is non-trivial), I think this one is also.
,
Jan 6 2018
,
Jan 31 2018
In what version is this expected? I still see the original issue manifesting in Version 64.0.3282.119 (Official Build) (64-bit)
,
Apr 5 2018
I don't see the issue anymore in Chrome 65. |
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by ligim...@chromium.org
, Oct 17 2017