Adding place=square name rendering

[kocio-pl]

Resolves #2203.

Rendering the same as other roads_area_text_name objects.

Warsaw, z16 - two squares which are not simple footway or pedestrian areas:
Before

After

[kocio-pl]

The question is how to render squares which are tagged as points? This is a valid tagging according to Wiki (some shapes would be hard to define) and it's about 10% of all the cases.

[imagico]

place=square currently does not have a language independent definition. It is currently defined as things that are named in a certain way in English, French, German, Italian, Spanish and Russian - which all mean at least slightly different things. The wikipedia article linked to from the wiki has a definition that is incompatible with the name based definition in OSM and practical use of the tag. I think this disqualifies it from being rendered here at the moment.

As said before i would be against text only rendering of areas based on way_area for features where there is usually no verifiable extent because that will likely lead to mappers applying a very wide interpretation of the extent of a particular square they consider important.

[kocio-pl]

I was thinking of square as a missing element of our road system and similar enough to simple pedestrian squares, but maybe moving it to placenames makes more sense technically.

I agree with you that definition on wiki should be improved (probably with something like "open space common in urban centres, like cities or towns"). For me however this is not a blocker for rendering this tag - currently parts of the complex squares are tagged with a name. Sometimes it makes sense (main road which changes name on square boundaries), but it's also a common workaround just to render it somehow (disconnected pedestrian areas or even parking aisles). I find it to be worse problem (breaking intuitive rule One feature, one OSM element) than vague square definition.

[Circeus]

In this specific example, it seems to be these are functionally parks (grass+trees+walkable), and there is no reason for the black rendering other than people not wanting a green-colored label on something without "park" in the name. Another way to look at it is that the example is not so much seeking to properly tag any sort of plaza so much as an obvious, roundabout (hah!) fix for issue #1253 on these two roundabouts. I.E. it's just tagging for the renderer either way.

I see no reason whatsoever not to define Plac Teatralny as highway=pedestrian area=yes. From the aerial imagery, the area is, as expected, a pedestrian area that happens to be bisected by a road, but to the best of my knowledge (a little limited, admittedly, especially where routing is concerned), there is nothing about that which would be considered improper for semantic or routing reasons as long s there are nodes where the road intersects the edge of the plaza (and now that I think of it, it would make Empire State Plaza look much simpler and nicer than the way I originally did it bu eliminating a lot of unnecessary footpaths).

Highway=pedestrian is defined for plazas and squares on both its page and "Map Features". The concern with "all objects on the square" on the place=plaza page seems more clearly related to a reluctance to use relationships than any tagging problems.

[kocio-pl]

I think you're missing the complexity of this square (I guess you talk about Plac Inwalidów): it consists of two "functional" parks, two one-way roads separated with a grass area, tram area and a road around it (BTW it's not a full circle, so not a roundabout) with footways.

[kocio-pl]

I see no reason whatsoever not to define Plac Teatralny as highway=pedestrian area=yes.

Only part of a square is pedestrian area (and as you can see it was tagged with a name once) - but there is also similar one (currently not drawn) on the front of the theater. The rest of the square is a parking space and roads, not pedestrian area.

[Circeus]

it consists of two "functional" parks, two one-way roads separated with a grass area, tram area and a road around it (BTW it's not a full circle, so not a roundabout) with footways.

There's the reluctance to use relations I was talking about! Take "Park" A and "Park" B (you call them "grass areas", I say they clearly function as parks), combine them into a relation and BINGO Plac Invalidow is showing.

Everything else is nothing but hair splitting.

• Don't derails this with semantic argument about the roundabout. You know exactly what I meant.
• The road are already name tagged, yet somehow you are not happy. Why? Because the result is no area tag, My solution (I think) should provide that. Whether or not that area specifically encompasses the (already named) surrounding road is irrelevant for all purposes anyway (same goes for whether the tag architecture explicitly include the crossing Adama Mickiewicza in the square).
• From the point of view of mapping this and getting the name to show, the ontological argument as to whether or not the trams are "part of" the Plac or just located at and named after it is completely irrelevant as it impacts neither the routing nor the display.
• You absolutely need those footpaths part of the Plac? adjust the park so it surrounds them instead of encompassing only the grass and tree. Duh. (Personally I wouldn't tag various bits the way you have, but I'm not gonna start an argument over that.)

A similar argument goes with Plac Teatralny, where you just happen to have drawn a lot of "enlarged sidewalks" as paths in this specific area, and rather inconsistently to boot, I should add. If they were all traced as pedestrian areas (which would render far more accurately the spots with sunken planters at each end of the parking, I should add), it would make a lot of sense to either just connect them all into a single pedestrian area (which would obviate the need to connect to the streets with footpaths. Win-win,) or into a multipolygon with your preferred of highway=pedestrian or highway=footpath that could get an area name.

Whether the streets or parking are formally part of Plac Teatralny, again, is irrelevant for the purpose of getting the name on the map (since the name and the exact precise area are not visually connected either way, even when using place=square!), and my way is semantically more accurate (it's not perfect, but a lot of OSM semantics aren't, and it's accurate enough in this case).

Another bonus of my suggestions is that routing engines can take you to Plac Teatralny right up (since it's the name of a pedestrian area or group thereof). Otherwise you have to wait and hope they will maybe eventually take place=square into account. If you're really anal, you can still use it on the relationship, but I would never do so for Plac Invalidow, which I don't consider to be a square in the first place, as I pointed out in my initial answer.

[dieterdreist]

sent from a phone

On 5. Jul 2017, at 10:22, Christoph Hormann ***@***.***> wrote: place=square currently does not have a language independent definition

as it is about a toponym I think it is correct to be language dependent.

[dieterdreist]

sent from a phone

On 5. Jul 2017, at 16:00, Circeus ***@***.***> wrote: The road are already name tagged, yet somehow you are not happy. Why?

because a square is not a road. Streets and squares together form the open space of settlements, honestly I find it disturbing we even have to discuss this here, I would have expected it is obvious to anyone interested in (urban) geography.

[Circeus]

because a square is not a road. Streets and squares together form the open space of settlements, honestly I find it disturbing we even have to discuss this here, I would have expected it is obvious to anyone interested in (urban) geography.

Thank you for derailing this with a tiny, irrelevant point that ignores my actual argument. The condescension is a nice touch, too.

[kocio-pl]

@Circeus: You've made a lot of interesting statements about how to tag squares. But it's hard to keep up since you expand already written entries, so I get lost eventually. More importantly - this is not the right place to do discuss square definition, let's keep on talking on Tagging list where it belongs. Currently relation is not allowed for squares on Wiki, but this tag needs more care anyway, so we can discuss also this point.

In this ticket I'd like to not check whether particular example is tagged as it should be, but rather how to achieve rendering both for points and areas tagged with place=square assuming the tagging was right.

[matthijsmelissen]

What is the status of this proposal?

[kocio-pl]

Waiting for definition fix. Wiki page should be probably changed to be clear that this is about toponym, not the land features. I'm currently working with easier problems, but plan to take care about wiki.

[matthijsmelissen]

Tagging as request changes, for administrative purposes.

[kocio-pl]

Just a note: discussion in #2816 might lead to some general conclusions about place=* tag, so I prefer to wait for the outcome.

[polarbearing]

7.5k now
Wiki definition improved.
I don't see a general conclusion about the place tag necessary to start rending this value. In case there is a general conclusion eventually, this case could be simply included.

[dieterdreist]

2017-10-30 23:38 GMT+01:00 polarbearing <[email protected]>:

7.5k now Wiki definition improved. I don't see a general conclusion about the place tag necessary to start rending this value. In case there is a general conclusion eventually, this case could be simply included

+1, please start rendering this, the tag was overdue, is rapidly gaining attraction even if not rendered, and is not disputed in any way.

[kocio-pl]

Now that the definition is fixed, I plan to do it soon.

[kocio-pl]

I have updated the code to omit parks and the likes, which are sometimes called a "square".

[polarbearing]

Hooray!
Just for understanding the code, that means when an object is tagged as park and as square, the label style for park is used?

[kocio-pl]

Yes, exactly.

[d3d9]

So why exactly did you decide to not render names for points?

[kocio-pl]

I don't remember the details, but that was certainly not a decision to avoid rendering points. Thanks for spotting it though - would you like to prepare the missing code?

[d3d9]

I have never done something like this before, I would have to try things out first.
How do you suggest doing it? Should it be a new layer in placenames.mss or can I make use of something existing in roads.mss or somewhere else? Or is there something obvious that I'm missing?

[kocio-pl]

I don't know exactly. My advise is to start with general technical description and setting testing environment first and then play a bit with a code:

https://wiki.openstreetmap.org/wiki/Standard_tile_layer
https://github.com/gravitystorm/openstreetmap-carto/blob/master/DOCKER.md

Let me know if you need some more help.

[polarbearing]

Does points mean place=square on nodes?

[kocio-pl]

Yes, this is what I understand.

[d3d9]

I tried this https://github.com/d3d9/openstreetmap-carto/commit/339b3b85cc7bd69d6ff57e9e76f6d4e3df15323e, how is it?

z17

z18

z19

Data (4 objects with place=square):

[kocio-pl]

I had no time to look closer, but on the first sight on the code and renderings it seems like a proper solution. Could you make a PR?