highway=proposed is too prominent, especially with tunnel=yes

[matkoniecz]

example of planned tunnel under mountain: http://www.openstreetmap.org/?mlat=50.0562&mlon=19.8919#map=15/50.0562/19.8919

[matthijsmelissen]

Do you think a change is necessary? We can of course not display all data we have in the database, and using to many different types of lines will not make the map easier to read.

If people think something like adding black lines around the road is a good idea, I can implement it.

[Rovastar]

Not sure of the need myself. Yet another combination for the key/legend. Construction is construction and it is (normally) a temporary measure

[matkoniecz]

1. It is not for construction, it is for proposed roads (construction on this one will not start before 2026)

2. I think that good solution is not displaying anything for proposed tunnels or making it highly transparent.

IMHO in general, highway=proposed should not appear or be way less noticeable (compare humanitarian layer: http://www.openstreetmap.org/?mlat=50.0562&mlon=19.8919#map=15/50.0562/19.8919&layers=H )

[sb12]

I think proposed roads should be generally shown less prominent. I can think of something like rendering only the outline of the road (transparency=80%) and have a very transparent filling (20%): Similar to how tunnels are rendered with wider gaps:

or maybe even like that:

The original rendering is here: http://www.openstreetmap.org/#map=19/49.14236/8.60837

This would also have the advantage, that you can better distinguish whether this is a proposed motorway or a proposed hiking path.

Not sure about how to show bridges and tunnels though: maybe by varying the darkness/transparency of the outline?

[HolgerJeromin]

Do we really have to render tunnels and bridges special for proposed? On construction we should, yes.
In the example of 2026 there could be changes of bridge/tunnel till construction.

[matkoniecz]

The problem is that proposed tunnels are rendered, with way too noticeable style. (exactly like normal proposed highways)

[matthijsmelissen]

I even wonder if proposed roads should be rendered at all on an online map. For paper maps, printing the proposed roads makes sense: the road might be constructed when the map is still in use. The same holds for cached offline maps, of course. But an online map can be updated easily, and the information of proposed roads is not much more useful than the information about roads that just have been destructed.

I suspect we might just be rendering proposed roads out of habits, without actually realizing what the reason for this habit is.

A principal question is also if we create a stylesheet for the openstreetmap.org map (that is not cached after road changes), or whether we want to generate tiles for services that do caching (or printing) as well.

[matkoniecz]

What worse, proposed roads are usually almost completely unverifiable.

And at least in my country (Poland) many proposed roads are nearly fictional (this one is proposed for years, always with planned construction starting later than 10 years into future).

Even with printing and I am unsure about value of presenting this data. Caching - I think that nobody caches this data for years.

Note: I changed ticket title to "highway=proposed is too prominent, especially with tunnel=yes".

[vincentdephily]

I think having proposed roads rendered is actually very usefull for a general map, because they are by definition still open for discussion and it is much easyer for the general public to check osm.org than to dig through pdfs on the local council's website. I have one such controversial project in my neighbourhood, and having an easy map to look at has helped the debate.

For "nearly fictional" roads, it's up to the mapper to decide wether a particular project is worth including in OSM.

That said, I do agree that the rendering of proposed roads is currently too prominent. I like sb12's second rendering.

[Rovastar]

Sorry I thought this was construction.

Nothing like mapping what is on the ground.
2026!

[matkoniecz]

I am unsure is it a good argument, but highway=proposed is the only known to me object displayed by the default layer but not included in JOSM presets.

[javbw]

+1 for a more transparent render.

[tyrasd]

Another thing is that proposed roads are rendered the same regardless of their classification. (e.g. this proposed cycleway bridge is rendered the same as a proposed motorway would be).

[sb12]

To solve the problem with the missing classification we need to have the "proposed" key in the database, which is apparently not the case at the moment. What is the procedure to have this key in the database?
My proposed rendering also only makes sense if the proposed key is available.

[Rovastar]

Adding additional keys is a more painful process. The whole backend will have to be changed.

I think in general we are avoid adding major changes to keys at this time. There is enough to do at the moment.

I would suggest the next stage of the roadmap will include features that require adding additional keys for use in osmcarto.

[pnorman]

I believe we're holding off any database schema/.style changes until 3.0. I'm working on some benchmarking that may establish an easier route which doesn't degrade performance, but don't expect anything quick - benchmarking takes lots of time.

[matthijsmelissen]

How would a change in converted tags work exactly? I suppose that implies writing a new osm2pgsql style? Do we try to update the default style, or do we create our own version? In the last case, we should perhaps include the style in this repository.
What are exactly the technical problems with importing the database again? Is it a computationally heavy operation? Are there any other programs consuming the postgis database on the openstreetmap.org servers apart from the Mapnik stylesheet?

PS: maybe you can find out what is the quickest way to benchmark? ;)

[matthijsmelissen]

@pnorman or @gravitystorm Any comment on my questions above?

And with respect to the original question, do you think we should keep rendering highway=proposed?

[gravitystorm]

@math1985 the osm2pgsql style file is already tracked in this repository - see https://github.com/gravitystorm/openstreetmap-carto/blob/master/openstreetmap-carto.style (albeit the contents of the first line needs changing now!). The problem with making changes is indeed that it's a big operation for anyone who is using these stylesheets. I aim to keep the openstreetmap-carto.style and mapnik versions stable as much as is possible, maybe change them every 6 months to a year or so at most.

As for highway=proposed, lets keep rendering it for now, but I'm 50/50 on whether it's something that a standard map style would be involved in. Certainly between construction, proposed and access rules we have a lot of confusing symbology.

[matthijsmelissen]

What about this question:

Are there any other programs consuming the postgis database on the openstreetmap.org servers apart from the Mapnik stylesheet?

If so, that would make things even more complex, because then we also need to sync with these other projects.

[gravitystorm]

No, there's no other programs doing this. And if there were, that would be for the osmf sysadmins to worry about!

But like I say, making changes to the osm2pgsql style file causes a lot of hassle, so it would take a good amount of coordination to make sure that it would be what we need for the next 12 months.

[pnorman]

I'm hoping to get some answers about hstore performance, but it's not trivial to rewrite all the queries and change the .style file to be more appropriate. hstore would make this easier

[Klumbumbus]

+1 from me to all of this comment http://github.com/gravitystorm/openstreetmap-carto/issues/345#issuecomment-35738221

[matthijsmelissen]

Closed by #1216.