On Fri, 24 Feb 2023 06:26:21 GMT, Karthik P K <k...@openjdk.org> wrote:
> > I think this class may benefit from a few tests that test with a very wide > > caret, to see if positioning is what you'd expect in those cases as well. I > > get the impression a lot of the code assumes a narrow caret (1 or 2 pixels) > > and this is why we see constants like `0` and `1` in the code, and even > > places where the caret width should be subtracted but isn't because it is > > assumed to be `0`. > > Since caret width is hard coded and there is no direct way of changing the > width, I believe we can proceed without adding tests with wide caret. Yeah, that's okay; it seems to make a lot of assumptions about the caret, even going as far as ignoring a caret with exactly 4 path elements (as apparently there is no way to ask if it is a split caret when RTL/LTR mixed text is encountered). I guess those will need to tackled if we ever change the caret shape. ------------- PR: https://git.openjdk.org/jfx/pull/980
