Torstein Hønsi

> Here is a stress test, 1000 points, a dataLabel for all of them, rotated differently: Are those just rotation, or are they text paths? Because those are two different...

Thanks! Here's a false positive then: https://jsfiddle.net/highcharts/7852fu9y/ . It probably falsely detects an overlap because text 2 is surrounded by text 1. I still think this would be avoided if...

Here's what I meant by part pushing, part unshifting the points: ```diff diff --git a/ts/Core/Renderer/SVG/SVGElement.ts b/ts/Core/Renderer/SVG/SVGElement.ts index 135d99f851..939f6fb575 100644 --- a/ts/Core/Renderer/SVG/SVGElement.ts +++ b/ts/Core/Renderer/SVG/SVGElement.ts @@ -1572,10 +1572,8 @@ class SVGElement implements...

> I wish I knew a cheap recipe for more stable polygons which could handle twisting paths better Yes, the current algorithm is increasingly wrong the more rotated the glyph...

@MarkusBarstad I see you ended up using the center of `getExtentOfChar` instead of `getStartPositionOfChar`. Did you try `getStartPositionOfChar`?

> Yes I tried it, @TorsteinHonsi , but I had some issues with it @MarkusBarstad Okay, something probably went wrong. In the latest commit I applied it, it looks good...

Thanks, we're closing in on the point when I can't tear this apart anymore. I found two fails now: 1. Text on a curve. This worked correctly before your last...

> To make a polygon surrounding the curved textPath in your fiddle, we have to increase number of character-positions read, right? I don't think so. It did work [after my...

Not sure if I understand the question, but by definition all texts with a break also have nested tspans. [Only the oneliners have raw text](https://github.com/highcharts/highcharts/blob/e8b2bc9858f3d2ddf46a1ab9060eb0147c4d59b1/ts/Core/Renderer/SVG/TextBuilder.ts#L121-L134).

Thanks! Remember to create unit tests as you go along though. Now that we're down to fixing edge cases, it's a high risk that one fix introduces failures in previous...