Fixed bug #3101, problem on ellipse2poly function#1139
Fixed bug #3101, problem on ellipse2poly function#1139opencv-pushbot merged 2 commits intoopencv:2.4from StevenPuttemans:bugfix_3101
Conversation
|
Looking at bugfix 3101 (http://code.opencv.org/issues/3101), Guanta also mentions changing the delta to float. Since noone explicitly confirmed this is correct, I did not yet applied the change. |
|
@vpisarev Vadim, could you please review this fix? Also, what do you think regarding changing delta to float? |
modules/core/src/drawing.cpp
Outdated
There was a problem hiding this comment.
Either missing braces or incorrect formatting here (this line shall be moved to the left)
There was a problem hiding this comment.
Hmm, let me check Guanta's entrence in the bugfix
There was a problem hiding this comment.
Missing brackets it is 👍 will adapt!
|
Any news on changing the delta value or not? |
|
@vpisarev , I do got your email notice but it doesnt show on github. Does your ok means that we could change the parameter to float? |
|
I saw today that a warning appeared: List failed tests: Anyone has a clue why this happens all of a sudden? Before the warning wasn't there. |
|
👍 |
|
And the error is gone again :) No idea why this happened :) Weird stuff going on on pullrequest website sometimes... |
|
👍 |
Added the fix of Guanta solving the following segmentation fault:
Before adding the points to draw to the output pts-vector it is resized to 0 (drawing.cpp, line 873). If now the ellipse to draw isn't valid, e.g. starting angle or ending angle are equal, then the loop won't be executed, no point will be added to the pts-vector and the line 894: pts.push_back(pts[0]); will result in a segfault.