Closed uliska closed 10 months ago
That sounds absolutely reasonable to me. However, l-zigzag-width and r-zigzag-width must be a divisor of the respective Y extents (which would now be increased by 2*border-width).
A solution would be to round the zigzag parameters to the nearest sensible value (e.g. in the original "leftZZspan" function).
I think the "reference point" should be the transition between fill and border. That also means the explicitly passed Y coordinate should refer to the inner edge of the border. So when
border-width
increases the border frame should grow to all four directions.