Bridge=Yes does not render unless associated with rendered highways/railways

[Rovastar]

A bridge by itself does not render in needs a rendered way for it to appear.

Related to #542

[dieterdreist]

2014-06-09 13:41 GMT+02:00 Rovastar [email protected]:

A bridge by itself does not render in needs a rendered way for it to
appear.

because "bridge" is an attribute, it is not defining a bridge by itself,
rather it states that something (e.g. road, railway) is ON a bridge. The
current rendering rules seem OK, if you want to tag a bridge, please use an
appropriate tag for it.

[matthijsmelissen]

Do we have a way to tag wild passages (not sure how they are called in English)?

[matkoniecz]

not sure how they are called in English

enwiki has it as "wildlife crossing" http://en.wikipedia.org/wiki/Wildlife_crossing

[Rovastar]

because "bridge" is an attribute, it is not defining a bridge by itself, rather it states that something (e.g. road, railway) is ON a bridge. The current rendering rules seem OK, if you want to tag a bridge, please use an appropriate tag for it.

I understand why but if there is a bridge should we render it? Probably yes, as it is an actual thing.
In theory you could have something that is difficult to define and not rendered as a way.

[matkoniecz]

IMHO disused but existing bridges and bridges with otherwise unrendered objects should appear (for example pipelines, wildlife crossings etc).

[matthijsmelissen]

Nice collection of examples: http://nl.wikipedia.org/wiki/Grote_nutteloze_werken
Strangely enough, the page is not available in English.

[matthijsmelissen]

So how do we tag things like this?

http://upload.wikimedia.org/wikipedia/commons/thumb/9/91/ViaducInutile.jpg/1024px-ViaducInutile.jpg

[matkoniecz]

So how do we tag things like this?
http://upload.wikimedia.org/wikipedia/commons/thumb/9/91/ViaducInutile.jpg/1024px-ViaducInutile.jpg

bridge=yes, highway=trunk(?), access=no, oneway=yes

[HolgerJeromin]

Wildlife crossing is wood over a tunnel for me.

[matthijsmelissen]

Also related to #436.

[vincentdephily]

Not sure if this is a good idea, but man_made=bridge could potentially be used as a physical thing that we can render ? I'm just afraid that it could lead to cargo-culting and tagging for the rendered. It also feels weird that mad_made=bridge would be needed for some objects but not others.

On the other hand, I'm not sure I agree with the "bridge=yes is just an atribute" claim. It seems rather pedantic and, even ignoring the issue of a bridge supporting something that we don't render, there is the case where a surveyor or amrchair mapper will see the obvious bridge but have no clue as to what it supports (if anything) and decide to only tag bridge=yes until he knows better.

[dieterdreist]

Am 12/giu/2014 um 19:09 schrieb vincentdephily [email protected]:

On the other hand, I'm not sure I agree with the "bridge=yes is just an atribute" claim. It seems rather pedantic

an explicit bridge object would also solve the longstanding problem of distinguishing several parallel bridges from one bridge with several carriageways (confirms the attribute point of view)

[vincentdephily]

an explicit bridge object would also solve the longstanding problem of distinguishing several parallel bridges from one bridge with several carriageways (confirms the attribute point of view)

Yes, that's what man_made=bridge and the corresponding relation are for. Go use them :) BTW, it would be great to get osm-carto to render it.

But you can't expect this tagging scheme to be used everywhere. Especially in the cases that osm-carto currently doesn't render, mappers are likely to use the simpler bridge=yes tagging scheme. Looking at the wiki, the wording never implies that bridge=yes is an attribute of the object rather than the object itself. Tagging bridge=yes on its own is a reasonable thing to do.

Maybe we need to change the rendering of briges yet again to be drawn around the ways rather than on the way's casing. That'd make the rendering consistent with multi-way bridges, and might make standalone bridges easyer to reason about.

[daganzdaanda]

Vincent, what do you mean with

That'd make the rendering consistent with multi-way bridges

The bridge relation?

[daganzdaanda]

There is a special kind of wildlife crossing bridge here: http://www.openstreetmap.org/way/199456384 and here: http://www.openstreetmap.org/way/198641856
These are bridges for bats :-) The bats seem to like to fly close to the ground, and since the road has been built, it would be very dangerous for a bat. If the bridges work, I don't know, but they are there, and are tagged correctly as far as I can see (man_made=bridge would still be better IMHO), and it would be nice to see them on the map.

I would say, let's render man_made=bridge and the bridge relation (#436). If it's an area, fill it with a grey colour. And if it's on a way, draw it similar to a bridge now.

[dieterdreist]

2014-06-13 12:17 GMT+02:00 vincentdephily [email protected]:

Looking at the wiki, the wording never implies that bridge=yes is an
attribute of the object rather than the object itself.

I think this is pretty clear from the actual mapping: you add it to the
road / rail and not to the bridge, hence adding the bridge implicitly. You
can improve the wiki if you like, it's a wiki ;-)

[dieterdreist]

2014-06-13 16:33 GMT+02:00 daganzdaanda [email protected]:

I would say, let's render man_made=bridge and the bridge relation (#436
#436). If
it's an area, fill it with a grey colour. And if it's on a way, draw it
similar to a bridge now.

+1
maybe exclude those man_made=bridge from rendering that have a railway=rail
or highway=* tag? Otherwise we would again be encouraging inconsistent
mapping (e.g. ambiguity where the name and other tags apply to).

[RussNelson]

This is only a problem because somebody decided that abandoned railways should not be rendered at any zoom level even if any non-rail-fan can see the bridge that they used to go over. E.g. these bridges (and one tunnel) are invisible in the standard OSM rendering:

The solution seems obvious to me: If it says "bridge=yes", then render whatever kind of highway or railway or waterway goes over the bridge, even if it would otherwise not be rendered.

[matkoniecz]

@RussNelson - can you link some mapped bridges with railway=abandoned? I plan on making pull request that would restore rendering of [railway=abandoned, bridge=!no].

https://www.openstreetmap.org/way/208510269 was already linked in #542

http://www.openstreetmap.org/way/4929040 is linked below

[matthijsmelissen]

can you link some mapped bridges with railway=abandoned?

http://www.openstreetmap.org/#map=16/52.4760/-1.8793

What do we do with abandoned railway tunnels that are now cycling paths? I.e. http://www.panoramio.com/photo/76193510. Note that the default is that railways render on top of roads.

[matkoniecz]

http://www.panoramio.com/photo/76193510

This one would be IMHO [highway=cycleway, tunnel=yes, lit=yes, surface=asphalt]. There is no trace of railway here (maybe in the name?), so there is no place for railway tag (note: I based this on the single photo, maybe there is reason for more tags).

[matthijsmelissen]

There is no trace of railway here (maybe in the name?)

I think @RussNelson would disagree. This is a typical example of an abandoned railway that is still visible in the landscape (the tunnel used to be a railway tunnel only).

What about this one? http://www.omroepbrabant.nl/Graphics/Image.aspx?object=News&type=large&image=De_man_sloeg_de_hand_aan_zichzelf_aan_het_Halve_Zolenpad_1114133.jpg
Also an abandoned railway, but now used as cycleway. Does it deserve special abandoned-railway-rendering?

[RussNelson]

I'm having a bit of trouble finding them in OpenStreetMap because they're not rendered anymore. But I think I remember where a few are. I'll get back to you.

[RussNelson]

Here's the URLs to the four pictures I posted, in order:
https://www.openstreetmap.org/way/158624393
https://www.openstreetmap.org/way/196297895
https://www.openstreetmap.org/way/82008834
https://www.openstreetmap.org/way/146575418

[pnorman]

What should a bridge look like if the base feature isn't being rendered?

[matkoniecz]

I think @RussNelson would disagree. This is a typical example of an abandoned railway that is still visible in the landscape (the tunnel used to be a railway tunnel only).

What about this one? http://www.omroepbrabant.nl/Graphics/Image.aspx?object=News&type=large&image=De_man_sloeg_de_hand_aan_zichzelf_aan_het_Halve_Zolenpad_1114133.jpg
Also an abandoned railway, but now used as cycleway. Does it deserve special abandoned-railway-rendering?

I am unsure whatever even railway tags are appropriate. Mapping tunnel, embankments, change of landcover etc certainly should be mapped - but "X used to be here" seems dubious, what about logical extending this and mapping buildings that no longer exists?

Both images provided by @math1985 IMHO depicts something that used to be railway infrastructure but it no longer is one. IMHO any tags that are used to describe railwayness of these structures should be ignored in rendering (and that is the current approach).

On the other hand images provided by @RussNelson in #612 (comment) depict infrastructure that IMHO should be rendered and currently is not (only second would be currently rendered as road tunnel).

What should a bridge look like if the base feature isn't being rendered?

Just bridge casings without anything inside? It can also make pipeline bridges visible (small part of #640), though I wonder whatever rendering man_made=bridge would be preferable (#436).

[RussNelson]

Well (of course) the base feature, in this case abandoned railways, SHOULD be rendered. The trouble is that most mappers, myself included, never anticipated the day when abandoned railways would be treated the same as dismantled railways (even though they aren't dismantled), and so weren't careful about tagging dismantled railways. Since railway=abandoned is now a mix of (the real) railway=abandoned and railway=dismantled, and some people don't want to see railway=dismantled running through their mall or housing development (reasonable) they have resorted to not rendering railway=abandoned (unreasonable).

But, really, that's a separate problem. Just render the bridge as if it still exists (which it does) for its most recently used purpose (which it still could be). There are over 300 railway=abandoned bridge=yes in New York State and as of a month or so ago, all of them disappeared from OSM. That's a shame.

[pnorman]

Well (of course) the base feature, in this case abandoned railways

This issue is about of bridges should be rendered where the base feature is not, not if abandoned railways should be. If you want to discuss that, there is probably an issue covering that, or create a new one if not.

[RussNelson]

I am unsure whatever even railway tags are appropriate. Mapping tunnel, embankments, change of landcover etc certainly should be mapped - but "X used to be here" seems dubious, what about logical extending this and mapping buildings that no longer exists?

Because that's not logical. If a building no longer exists, there is no evidence that it was ever there and should not be rendered. If a railway no longer exists, there is almost always evidence that it was there, if only because the railway you can see has a gap in it. E.g. in the case of the bridge to the east of the tunnel (the third URL I posted above), it has rails to the east and west of it (in the tunnel), but no rails on it.

[dieterdreist]

Il giorno 28/giu/2014, alle ore 05:08, Russ Nelson [email protected] ha scritto:

The solution seems obvious to me: If it says "bridge=yes", then render whatever kind of highway or railway or waterway goes over the bridge, even if it would otherwise not be rendered.

what about tagging man_made=bridge or building=bridge? (and rendering it)

[matkoniecz]

see also https://trac.openstreetmap.org/ticket/3852

[systemed]

Further example that just tripped me up: this big-assed bridge is now not rendering due to having a railway=abandoned on top. OSM object: http://www.openstreetmap.org/way/70125797

[dieterdreist]

2014-09-01 12:17 GMT+02:00 Richard Fairhurst [email protected]:

Further example that just tripped me up: this
http://www.geograph.org.uk/photo/2787255 big-assed bridge is now not
rendering due to having a railway=abandoned on top. OSM object:
http://www.openstreetmap.org/way/70125797

we should really start mapping bridges generally, without relying on ways
passing over them, e.g. man_made=bridge which is currently used 1675 times
and defined in the wiki:
http://taginfo.openstreetmap.org/tags/man_made=bridge
http://wiki.openstreetmap.org/wiki/Tag%3Aman_made%3Dbridge

[Jonobennett]

Another UK example: https://www.openstreetmap.org/way/37548602 - while people do walk along the top of this, I don't think there's a legal right of way.

[kocio-pl]

Does osm-carto code contributors plan to deploy this rendering?

Taginfo says we have now 2 341 965 objects tagged "bridge=yes" (even the second kv pair, "bridge=viaduct", is notable - 41 583 objects), so the evidence that this feature is important is strong:

http://taginfo.openstreetmap.org/tags/bridge=yes

[dieterdreist]

2015-02-11 14:36 GMT+01:00 kocio-pl [email protected]:

Taginfo says we have now 2 341 965 objects tagged "bridge=yes" (even the
second kv pair, "bridge=viaduct", is notable - 41 583 objects), so the
evidence that this feature is important is strong:

http://taginfo.openstreetmap.org/tags/bridge=yes

if you read the contributions in this issue you'll find that "bridge=yes"
is not a bridge by itself, it is an attribute that is stating that
something else (e.g. a highway) is on a bridge, i.e. this is only an
implicit way to map a bridge.

[kocio-pl]

Sure - I just realized it later... =}

It means we don't have clear numbers, but still some of the bridges have multiple highways on it (those are probably important by definition and I guess not all of them have the outline tagged) and even those with just one lane can be better shown at maximum zoom level when rendered with the outline.

[daganzdaanda]

I think this issue should be closed, since it's misleading.
A bridge without a way should be tagged as a closed way with man_made=bridge, or as a bridge relation. Issue #436 deals with how it could look.

[systemed]

@daganzdaanda No, this issue is for bridges without rendered ways, not bridges without ways of any sort. For example, this modest structure: http://weirdnj.com/stories/paulinskill-viaduct/

[daganzdaanda]

OK, thanks, I got confused!
Yes, that nice bridge is correct as a abandoned railway with bridge=yes. But I'd still add the outlines with man_made=bridge, to be complete. Of course, neither is rendered yet...

[RussNelson]

We're not rendering the Paulinskill Viaduct? It dominates the landscape in Hainsburg. There is no place in town where you can't see it .... except if you're looking at OpenStreetMap.

I'm sympathetic to the idea that if you can't see it, it shouldn't be rendered. Can I get a little sympathy for the idea that if you CAN see it, it SHOULD be rendered?

I'm also sympathetic to the idea that rendering every railway=abandoned is wrong, because there are places where even I cannot see that a railway used to go. So how about this suggestion: we render railway=railbed, for those places where an ordinary mapper can see a railbed, whether on aerials or on site. That way, those of us who can see the Paulinskill Viaduct can tag it as railway=railbed, and we can close this issue once and for all.

[dieterdreist]

2015-02-17 10:57 GMT+01:00 kocio-pl [email protected]:

It means we don't have clear numbers, but still some of the bridges have
multiple highways on it (those are probably important by definition and I
guess not all of them have the outline tagged) and even those with just one
lane can be better shown at maximum zoom level when rendered with the
outline.

yes, I am absolutely in favor of tagging/mapping bridges as their own
objects, you also need this for other stuff like wikipedia links, bridge
names (esp. when the street over it has a different name), bridge refs, etc.
The "solutions" to this problem are just silly (bridge_name etc.) if you
can have a simple and logical solution: a bridge object and "name".
Basically these are workarounds that evolved around the common
misunderstanding that bridge=yes would be the way for mapping a bridge
itself and not just an attribute to say: way on a bridge.

[daganzdaanda]

We're not rendering the Paulinskill Viaduct?

https://www.openstreetmap.org/way/95507414
It's actually a railway=proposed since Feb 2014, probably the time that railway=abandoned stopped being shown.
I just added an outline with man_made=bridge and selected info from the wikipedia article:
https://www.openstreetmap.org/way/328801035

I'll change the railway on the bridge to "railway=railbed" since I think this is a good tag and I don't believe that there is really a current proposal for a new rail on that route. I'll not touch the rest of the railway at the moment, since I don't know where exactly "abandoned" or "railbed" would be best.

[SomeoneElseOSM]

What - we're now proposing tagging changes from OSM carto github issues now?

http://taginfo.openstreetmap.org/search?q=railway%3Drailbed

(currently 0 uses)

Well, I suppose it makes a change from telling the style developers how to render stuff from Every Other OSM List :)

[daganzdaanda]

Oops ;-)
I thought that Russ had a proposal for his railway=railbed idea a few months ago??
Well... back to "abandoned" for the time being.

[RussNelson]

Yes, I did propose it, and while it seems perfectly sensible, and a solution to everyone's objections in every regard, I got no positive responses. Do I need to make this suggestion in the form of a pull request?

Yes, nobody is using railway=railbed now. That's the whole point -- people don't want railway=abandoned to be rendered because it is probably 30-40% MISUSED. Yet rendering encourages mapping, so if railway=railbed starts getting rendered, then people will be rewarded for fixing all the wrong railway=abandoned's.

[pnorman]

Please remember what the issue is about. It is about ways with bridge tags with no other tags that are rendered, it is not specific to railways. This issue would not change if there were changes to what railway tags were rendered.

If you believe there are new tags that should be rendered create a new issue for them, do not discuss them here.

[RussNelson]

Please remember that this problem was created by ceasing the rendering of abandoned railroads. If we can fix that problem, this problem will be fixed at the same time.

If you believe there are new tags that should be rendered which will solve this problem, please discuss them here.

[pnorman]

Locking until @gravitystorm gets back. This issue has gone entirely off-topic, and no discussion is being done on how a bridge should look if its underlying feature is not rendered.

[gravitystorm]

This issue is closed since it has been driven far off topic.