Natural=bay multipolygon or single dot dispute

Continuing the discussion from Cursed Ajaccio Gulf:


Three years ago, I replaced in Corsica the natural=bay multipolygons by nodes after a fair discussion.
Alas, two years later, the predecessor editor restaured its multipolygon, masking again many natural=barerock islands that give meaning to the place names of the coast, the ones that are not mapped doubly as coastline.
Golfe d’ajaccio
I have no answer from him to my messages.
What do you advise me to do?

There is no reason imho to start a new topic for exactly the same issue as the one you want to continue here, so I would suggest you delete this post here and just add it to the Cursed Ajaccio Gulf topic to reopen the dispute again.

You are mentioning messages you sent to Olyon but there is no changeset comment for the CS when Olyon replaced the bay node with a multipolygon again. I would suggest you start with such a changeset comment and draw Olyons attention to the content of the original topic where he/she was not engaged in as far as I can see.

I agree with you that mapping a bay as (multi)polygon is not a good idea and it is not encouraged by the wiki, but Olyon is a very active mapper so he/she may have certain reasons to do so and the best place to discuss about it would be the original topic “Cursed Ajaccio Gulf” imho.

1 Like

I suspect that a new topic was opened because the old thread was closed. I don’t see a reply option there, at least.

Comment on the changeset. Ask them why they just did a revert without discussion. Also add a comment to your changeset discussion explained where the deletion of these monster multipolygons was agreed by the community first.

If all else fails, email the DWG**.

** I’m a member, but don’t believe that I have been involved in this particular gulf issue.

I will do so.
Thank you very much.
(Indeed the first topic is closed)

Good onya mate … that is correct and I did not notice it … :sleepy: … but why would someone want to close a topic like this one? Is there any rule which I have not noticed so far? I’m not arguing, just being interested.

I suspect that what is happening is that old threads are closed automatically after a while. This definitely makes some sense to me; on the OSM help site (where I’m a mod) a perennial problem is people resurrecting old threads and giving them a new prominence that they no longer deserve. What might have been a good answer in (say) 2013 may not be now. I’ve certainly had to ask people on IRC to vote down old “help” answers of mine that simply aren’t valid any more!

The “Cursed Ajaccio Gulf” thread is from 2020, so not exactly ancient history, but also quite a while ago. Presumably there is a site setting somewhere; maybe ask over in “this site feedback”.

1 Like

I believe this is because the openstreetmap category has been removed, thus prohibiting further discussion as part of it’s topics.

I have commented on the change and received a reply from the author. He is in denial about the consequences of polygons.
I quote him by translating into English

The multipolygon natural=bay has nothing to do with your problem, it only adds the name=* which is displayed without having to zoom in very much, unlike the node.

All this leaves us orphaned from our coastal rocks and it’s very sad! I feel it as a bullying to Corsica.

OSM wiki say that the key:natural should not be used on relations. Multipolygons are relations: it is clear, no?

That’s utter bobbins.

The “natural” key is very often used on relations. Woods with holes in and lakes with islands are common examples.

Edit: I’ve changed this. The wiki is confused (quelle surprise) because it talks about “nodes”, “ways”, “areas” and “relations” rather than “points”, “lines”, “polygons” and “things with no geometry”. covers all relations, which can have various different sorts of geometry or none, and the first example given there is “multipolygon”, which obviously does have a geometry. It’s pretty meaningless to tag something with no geometry as “natural=wood”, for example, but a multipolygon “natural=wood” makes perfect sense. If someone wants to edit that page to say that “it makes no sense to use the ‘natural’ key on objects with no geometry” that would help further.

Okay, but this is not the same case for several reasons.

  • The complexity is much less.
  • It is handled by a single cartographer.
  • The rendering of the sea is very slow.

In the natural=gulf case the edge effects are destructive for the work of other mappers.

I answered on the changeset in French. Here are the arguments I give:

2 multipolygons had been replaced by a node but also by a way that would close the bay. I don’t think anyone on a forum said to replace the bays with this:

I forgot to remove the tags on this one…

The multipolygon natural=bay have nothing to do with these problems of rocks not visible on the rendering, it only adds the name=* which is displayed at a zoom adapted to the size of the multipolygon unlike the node which can represent a bay a few meters or several hundred km.

An example of natural=bare_rock that does not appear on the render and is not part of a bay:

If they were present on the rendering before and they are no longer present, there may have been a modification in the rendering, the blue of the sea which is added over the rendering of the bare_rock and before it was the opposite, maybe…
Rendering is manage here, if you want to ask:

The other solution is to surround the islets with a natural=coastline. But natural=coastline means that the islets are almost always above sea level.
If you do that, you will have to be patient, updating the “blue” of the sea sometimes took several weeks or months.

The islets need to be natural=coastline to appears on the sea.
The natural=coastline islets does not appears on the see if there are inside a natural=bay polygon but not part of the relation.
Inserting hundred islets in the relation is too complex, not done and so islets are erased.
The islets have names and/or explain the names of the capes and bays. They are needed to give sense to the map. What did Corsica and Sardinia do to you to make you want to erase their coastline?

On the “OSM Carto” renderer I suspect that it might, because of the way that sea polygons are handled. Compare here:

with here:

The first of those (OSM Carto) draws sea polygons over tidal rocks, the second of those does not.

It is the problem : I wish to do that! But it will not work! The reason will be that the islets will not be part of the natural=gulf polygon relation.

1 Like

Note that infobox has “relation”, “way”, “point”, “area” as separate entries

“area” covers closed ways and multipolygons, relation covers nonmultipolygon relations

(If I would be emperor of OSM then infobox would not have this fields, confusion is greater than usefulness)

You are so convinced that these relations are a problem that you do not try to understand what you are being told…
How do you explain that this rock Way: 664753630 | OpenStreetMap is not visible on the OSM Carto renderer? There is no relation natural=bay here. The rock next to it Way: 664037971 | OpenStreetMap is visible because there is natural=coastline Way: 4843142 | OpenStreetMap.
I can give other examples without bay that are invisible : le way 669471392, 504644564…
there are probably thousands of them…

… In which case it shouldn’t link to the general relation page which has multipolygon as the first example relation type!

good point, posted on Talk:Wiki - OpenStreetMap Wiki and hopefully someome™ will handle it.

1 Like

It is a dead lock loop !

1 Like