I noticed a very special edge case of using canvas(debug: true) together with fills in merge-path and some relative coordinates. I did not perform any further testing to minimize the repro code, so it is possible that the bug can be triggered in more general situations.
When activating the canvas debug plots in the following code, the fill areas of the triangle are inverted:
Without debug, expected form:
With debug: true, unexpected fill area inversion:
Maybe I am also abusing the merge-path function here since the rect line draw order is not specified in the manual?
My goal was to avoid the fine white line between the rect and the triangle, which happens without merge-path in some PDF viewers.
I noticed a very special edge case of using
canvas(debug: true)
together with fills inmerge-path
and some relative coordinates. I did not perform any further testing to minimize the repro code, so it is possible that the bug can be triggered in more general situations.When activating the canvas debug plots in the following code, the fill areas of the triangle are inverted:
Without debug, expected form:
With ![image](https://github.com/cetz-package/cetz/assets/15750438/5458ad92-3770-4430-8dd0-f6954415f83c)
debug: true
, unexpected fill area inversion:Maybe I am also abusing the
merge-path
function here since the rect line draw order is not specified in the manual? My goal was to avoid the fine white line between the rect and the triangle, which happens withoutmerge-path
in some PDF viewers.