[dsign]

> I’d say the takeaway is that treating everything like a black box doesn’t produce good results (sometimes). And not knowing enough to open the box completely leaves you with a lot of unknowns, surprises, and loose ends. But hey, figuring this stuff out is why we get paid the $$$.

To add a little bit of nuance to your comment, I've to say that computational geometry algorithms are incredibly human-effort-intensive. Treating their implementations as a black-box could be an incredible time-saver. In my experience however, that particular black box tends to fail when least one needs it, and then one is left with inscrutable pieces. Often the "least bad" approach is to learn how polygon/polyline composition works and implement it oneself. It will be buggy and crappy (at first) but it won't break in cryptic ways. Budget a couple of months for the learning and initial implementation though, and be most wary of floating point and God coming in the night to mess with your code and put new bugs.