Comment 9 for bug 722544

Interesting points – I was mostly fighting the feeling that the "intuitive pick" for a angle domain was ]-180, 180], decided for myself that the win when mapping the [-180, -100] range to [180, 260] was worth it despite moving out of the "more intuitive" clockwise or counter-clock-wise rotations (and choosing, slightly arbitrarily, that -90 was a better cut-off point than -10, for human-intuition reasons). :)

There are certainly arguments many ways possible around this; some of which might be best inspected by taking a large sample of svg:s in the wild and running some stats on what's commonly used / generated by commonly used svg-generating tools. But that's another ballpark of work than coming up with clever optimizations. :)

Scour is fun!