mark padgham

no worries - ping me any time, and all cool about (non-)PR, especially since i haven't even looked much into your code, and don't imagine finding much time any time...

BottleRocket2 is in your git version, as are two bottlerocket images. However, neither of the images is formatted like the others (and "./figure/bottlerocket-1.png" is much larger), so these images will...

That's a cool idea, and ... (my standard response by now) ... easy to do with `dodgr`. Just convert to a dodgr network, make frequency tables for `from` and `to`...

That's a really good idea. I would suggest using either a threshold *proportion* - so merge the shortest X% of segments, or a maximal threshold of numbers of segments, giving...

It is definitely fast and powerful, but comes with an important caveat: It it not a dual-weighted algorithm, so routing can only be done via actual geometrical shortest paths, and...

No, you can't unfortunately use a single-weighted Dijkstra at all for dual-weighted problems. Dual-weighted algorithms choose a path based on one weight, but calculate the resultant distance with the other....

Depends what you mean by "areas"? The new `dodgr` function, [`dodgr_fundamental_cycles`](https://atfutures.github.io/dodgr/reference/dodgr_fundamental_cycles.html) enables all areas *enclosed* by cycles of highways to be identified. Or do you mean something else?

Yeah, as @tim-salabim well knows, this all boils down to `osmdata` "pretending" to return "standard" `sf` objects, yet actually naming both (1) all list items; and (2) all rows within...

Have a look at the great code by @mdsumner in [`mapscanner`](https://github.com/mpadge/mapscanner/blob/master/R/aggregate_polys.R) that does the exact opposite of that - aggregates overlapping polygons, and returns contours for degrees of overlap. That...

Pinging @mdsumner for an answer to that So question...