I did some more testing on using the add-pois-to-lines and mkgmap:from-node: and got a little confused.

As far as I understand is the classic implementation of add-pois-to-lines to create a poi for every node on the line.

By checking for example mkgmap:line2poitype = mid you decide to use it or not.

Nice improvement on performance is the mentioned options to not create useless poi's and the idea to do it for start,mid,end,tagged or all

The manual states that all values of the line are copied to the node.

When testing for highway=turning circle it worked even too well as also the node-tags highway = turning_circle was overwritten by the highway = track value of the way it was part of.

Then Gerd started a 'branch' (if that’s the correct word) with the mkgmap:from-node: feature, and also this worked perfectly

Yesterday I started a new usage: There is a node barrier = gate on a highway = path where the barrier does not have access tags.

In my style the default for barriers without an access tag is to not allow access (why would they otherwise be added) and highways without access tags to do allow access (that’s where highways for)

But as sometimes forgotten by taggers to finetune access, I'd like to check if the barrier is on a highway with an official bike/hike route as set in the relation file.

In that case I like to add foot = yes or bicycle = yes. Always worked for the highways but because not set on the barrier-node the routing breaks.

I expected to be able to catch the barrier with

Osmid() = 2486838092

Check, catched but no more tags then barrier = gate

Current implementation r4521 (not the branch)

barrier = gate & highway = path & route = foot {echotagss 'all'}

but that failed (add-pois-to-lines ~ true)

Then switched to the new feature hopefully coming up in a future version (branch mkgmap-r4504.node)

mkgmap:from-node:barrier = gate {echotagss 'all'}

but that failed as well

Strange because in the turning_circle situations I even got 2 poi's. One for the node and one for being part of a highway.

My example where routing broke is a barrier = gate on a highway = path.

In this case only the osm() node seems to be created and not the "add-pois-to-lines node".

On the other hand (strange?) the highway = path does contain the tag: mkgmap:way-has-pois=true to indicate that the highway has barriers

Finally ...

Gerd launched the idea to create the tag mkgmap:copied-as-poi=true

If there are two pois created "the osm node barrier = gate" and the "add-pois-to-lines node" version, I only can use the "add-pois-to-line version" because that one gives me access to the line where it was derived from.

But this also means that when processing the "osm node version" it must be ignored only if later on, another one comes along. How do you know?

If a single stand alone barrier comes along, you do like to use it the same way as it is always has be done

Questions 😊

Can I expect to have the barrier = gate nodes always to created twice if attached to a highway and add-pois-to-line = true?

Can I catch it with the future functionality we started for the turning circle as well?

Do we best extend the mkgmap:copied-as-poi=true to be available both from the osm node version of the poi as well as the created poi?

I'm happy do more testing

Hopefully you guys can still handle my stories.

Kind regards, Joris

Test area

Node: https://www.openstreetmap.org/node/2486838092

(already updated to foot=yes, but not important for the test)

Highway https://www.openstreetmap.org/way/241008394

Walking route: https://www.openstreetmap.org/relation/3251904

To be sure it was not a baecamp problem

When only adding mkgmap:foot=yes to the node barrier=gate its oke

-----Oorspronkelijk bericht-----

Van: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> Namens Mike Baggaley

Verzonden: dinsdag 9 juni 2020 00:14

Aan: 'Development list for mkgmap' <mkgmap-dev@lists.mkgmap.org.uk>

Onderwerp: [mkgmap-dev] add-pois-to-lines documentation

Hi Gerd, please find attached a proposed patch to the documentation for the add-pois-to-lines option as requested.

Cheers,

Mike

>I've implemented the suggested improvements with r4507. Please suggest

improvements for the documentation changes.