Explaining a comment

[dbuenzli]

Hello,

Not an issue per se but the explanation of a comment.

I was reading your implementation and saw this comment here. The implementation of this part of the algorithm is quite convoluted in the reference implementation (available here which I believe this is a mirror on github). I haven't fully wrapped my head around yet…

The reference implementation uses indices 2 and 1 here. But if you are here in the code, left points coincide, this means you went through this line which pushes a 0 (NULL) at index 0, before the events.

In your code you track this via a boolean, not a NULL value in the events array. So this is why 1 and 0 are the correct values in your implementation, your events are not shifted to position 2 and 1 by a NULL value. :-)

[w8r]

Wow thanks a lot, this can be priceless! I am reworking the lib in typescript at the moment, I will double check this!! Kind regards, Alexander Milevski

…

On 19 Jul 2022, at 19:49, Daniel Bünzli ***@***.***> wrote:  Hello, Not an issue per se but the explanation of a comment. I was reading your implementation and saw this comment here. The implementation of this part of the algorithm is quite convoluted in the reference implementation (available here which I believe this is a mirror on github). I haven't fully wrapped my head around yet… The reference implementation uses indices 2 and 1 here. But if you are here in the code, left points coincide, this means you went through this line which pushes a 0 (NULL) at index 0, before the events. In your code you track this via a boolean, not a NULL value in the events array. So this is why 1 and 0 are the correct values in your implementation, your events are not shifted to position 2 and 1 by a NULL value. :-) — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.

[dbuenzli]

Cool.

Btw. one thing I noticed is that both the reference implementation and this implementation do not compute the topological information for the result (holes + contour winding) when the operation hits one of the trivial paths.

I guess this may surprise users of the API which may expect to have that in the result.