[BUG] TransportationName: calculate rank doesn't handle major roads that are not part of a preconfigured network

[onthegomap]

[BUG] TransportationName: calculate rank doesn't handle major roads that are not part of a preconfigured network

[boldtrn]

Sounds good, thanks for your work @msbarry 👍

[boldtrn]

Haven't heard back, is this something you would like a contribution? Or is this something that wouldn't be that relevant to you?

[msbarry]

Hey sorry I just haven't had a chance to look into this. For B313 at least it looks like it does show up (because each of the ways in it also has a ref tag), but only starting at z12:

The reason it doesn't show up at z11 and below is that openmaptiles sets a min length of 8km for transportation_name segments to appear at z11. The trunk portion of B13 is just over 9km so gets emitted, but planetiler translates that length limit to a min pixel size of the feature on a tile and that segment happens to cross a tile boundary

I can see two fixes here:

1. Use a larger buffer for transportation_name tiles at z11 and below so the min length computation includes the full segment
2. Then maybe to catch the case where a way is in a relation that has a ref but no network (and there's no ref on the way) we could add an option that breaks from openmaptiles compatibility and uses that ref on the transportation_name linestring

What do you think @boldtrn ?

[msbarry]

Use a larger buffer for transportation_name tiles at z11 and below so the min length computation includes the full segment

Actually we already do that... it turns out that the issue here is that some onramps/offramps to B313 are tagged with ref=B 313 (for example https://www.openstreetmap.org/way/845773628) and the line merging code stops merging at 3-way intersections and none of the line segments ends up exceeding the 8km min-lenth limit. I'm planning to refactor the line merging code at some point (#54) so it's more robust to these kinds of issues.

In the meantime you can run planetiler with --transportation-name-limit-merge which groups onramps separately from main roads. When I run with that option, I can see B313 shield as low as z8

That option is disabled by default since it differs from openmaptiles, but if you find that the output is actually closer to openmaptiles with it enabled then we can enable it by default.

[boldtrn]

Thanks @msbarry for the proposal. Are there any side-effects when using --transportation-name-limit-merge? Is there anything I should look out for when using this parameter?

[msbarry]

Nope, it should be isolated to only affect transportation_name layer, and in general will make road labels show up a bit more frequently on the map at lower zooms. It should not be a huge difference though.

[boldtrn]

Thanks @msbarry this parameter seems to indeed improve the rendering of highway signs. Compared to OMT there are still fewer highway signs (B10 and A8 are still missing).

Is there be something else that I could try? If a code change is required I can create a PR as well, but I might need a hint or two :).

[msbarry]

Are you sure they're not showing up in the underlying tile data and it's not a style rendering issue? When I look in tileserver-gl I see B10 and A8:

[boldtrn]

Thanks for the hint, you are right, sorry for the troubles, they indeed appear in the data (I should have checked that).

Should we close this issue then?

[msbarry]

@boldtrn I think this is good to close for now? #298 tracks the long-term improvements to feature merging