will anything be ever done about colinear edges restriction

[makc]

I mean, now I have to wiggle the vertices by 1e-something * Math.random () which I should not have to. Just have divisions by 0 replaced by some small number, and it should work.

[r3mi]

This is an issue that is present in the original algorithm, from the C++ version (now here : https://github.com/greenm01/poly2tri ).
You are suggesting ignoring COLLINEAR detection, and ignoring potential math error resulting from it ? Maybe it could work.
Are you able to suggest a patch ?
If not, I may have a look latter.

[makc]

Are you able to suggest a patch ?

I am not an expert here. What I can say is that this hack allows this test to pass:

[makc]

other tests like big roundish triangle and butterfly-like polygon fail, so the hack is no good.

[joel-simon]

Any update here?

[r3mi]

Sorry, not for the time being.

[obidobi]

What is the issue, putting steiner points on the polygon edge should split the edge?

Steiner points are supposed to be placed inside the polygon. I guess you will have to refine your polygon edges before sending it to the triangulator.