Bounds checking for get_cursor_data(). Closes #4957 #4961
Conversation
|
Less than thrilled about the the variability of the types, but I think it is ok in this case. I am more concerned that |
|
Ah, I see a bug there, I think. |
|
errr, I mean left/right/bottom/top |
|
It looks worse that that. In the case of the default extent it looks like |
|
Joys! Well, the fix is fairly straight-forward, if tedious. Just do a flip of xmin/xmax and ymin/ymax whenever needed. Stay tuned... |
|
I am thinking that is a red herring. The only place where doing an order check would have helped is with FigureImage's contain(), and its get_extent() operates in a way that always returns left/right/bottom/top. In any case, bounds-checking should be par for course, even if get_cursor_data() should not have been called in the first place. We could open up a new issue to investigate that further. Maybe a faulty |
FIX: Bounds checking for get_cursor_data(). Closes #4957
Decided to go with reporting
Nonewhen outside the image. There is still an oddity I discovered, and I highlighted it in the test with commented out code. With origin as "upper", you can go off the top-left corner of the image for a bit with a value being reported. This doesn't happen at any corner for origin "lower", though.