Style area-pins differently

[tordans]

Description

I was quite confused just now when I tried to select this bicycle pin inside the area:

It turns out, this is the pin that symbolizes the area, not a separate pin.

I think those should be styled differently, to make it easier to understand their meaning.

Btw: The placement is also a bit weird, I was expecting it in the center or maybe as a pattern, but not where it is ATM. When edit the shape, the pin moves around quite far, which was also weird.

Area: https://www.openstreetmap.org/way/718205697#map=19/52.48309/13.44155&layers=N

[bhousel]

Oh yeah this is a duplicate of #958 - can also look at #534 also for examples of where this rendering is better than what Rapid v1 and iD used to do.

We didn't really talk about it much there, but I'm kind of weakly against the idea of making the virtual points look different. If people really insist on it, I'd probably change the pin somehow, but then users will open issues to ask "why is this point different" and we'd have to explain them about areas and it's still not clear to me why a mapper would care about this detail.

If I'm trying to edit a POI, I shouldn't need to think about whether that POI is mapped as a point or as an area - this is really just a quirk of OSM. And it is really useful to be able to see and edit these mapped features at lower zooms where the area polygons are tiny.

Btw: The placement is also a bit weird, I was expecting it in the center or maybe as a pattern, but not where it is ATM. When edit the shape, the pin moves around quite far, which was also weird.

I agree in the screenshot the placement of the pin is weird. We use the Polylabel library to calculate the "pole of inaccessibility" see comment here. I don't totally understand the algorithm, but it is supposed to put the point somewhere within the large part of the shape, even if it's an 'L' shape or has holes. When you edit the shape it will adjust the point around to different places - this is expected.

[bhousel]

I agree in the screenshot the placement of the pin is weird. We use the Polylabel library to calculate the "pole of inaccessibility" see comment here. I don't totally understand the algorithm, but it is supposed to put the point somewhere within the large part of the shape, even if it's an 'L' shape or has holes. When you edit the shape it will adjust the point around to different places - this is expected.

Quick update: I did investigate the placement a bit - https://rapideditor.org/edit#map=21.82/52.48309/13.44156&id=w718205697

It looks like this issue is also mapbox/polylabel#33

• If you square the shape with Q a few times, it will move the point more towards the center.
• This means that the long rectangular shape actually starts out slightly trapezoidal, and the point wants to gravitate towards the slighty wider end.
• maybe we'll switch to one of the Polylabel forks mentioned on Gravitate to Centroid mapbox/polylabel#63

[tordans]

…but I'm kind of weakly against the idea of making the virtual points look different. If people really insist on it, I'd probably change the pin somehow, but then users will open issues to ask "why is this point different" and we'd have to explain them about areas and it's still not clear to me why a mapper would care about this detail.

I think the way iD handels this case is good: It shows the type but there is no confusion between the "pin-icon in area" vs. "area with type-icon" case.

iD	Rapid

This styling did not come up in any issues that I remember on the iD side.

[bhousel]

This styling did not come up in any issues that I remember on the iD side.

The old area styling has been causing issues in iD going back years.

• the position is at the centroid, which sometimes places it outside of the shape
• the icon disappears when the shape is small or zoomed out
• the icon is not clickable

All issues that have been raised in the past.