Talk:Editing Standards and Conventions

From OpenStreetMap Wiki
Jump to navigation Jump to search

Discuss Editing Standards and Conventions page here:


Cleanup this page

This Editing Standards and Conventions page is a bit of a confused jumble of very specific editing/tagging advice, out-of-date applet screenshots, people asking questions, and people answering them with personal opinions!

We need to decide what we want to achieve with this page.

I'm sure we don't want unanswered questions on the page there. They belong on this dicussion page. Even where the questions are answered, I suspect many of these do not reflect concensus. In fact many of them are signed. These also belong here as discussion.

Beyond that, there's a question of how this fits in with other pages such as specific pages found under Category:Keys and Feature Index

-- Harry Wood 10:20, 30 October 2007 (UTC)

Nothing really tackling this in the 7 years since I wrote that comment! I mean I think it may have improved a bit in that time in that there is some attempt to list more general things like tagging at the top of the page, but the page is still a jumble of different ideas.
Since then we've also established Good practice as a more heavily linked page which sets out a range of mapping principles. Kind of similar page idea.
A point I raised in the development of that page, applies here too. We should be able to link off to places elsewhere on the wiki to give more detail. So for example we have quite a large illustrated section on "divided highways". Perhaps that could be moved to its own page, as a useful thing to link to from pages like Tag:highway=trunk, and from help questions like this. Then on this page we carry a condensed three or four sentences summary and a link (a layout we are almost achieving on Good practice)
Ultimately then we could aim to merge this with Good practice
-- Harry Wood (talk) 11:35, 10 May 2015 (UTC)
I support your idea to merge this page with Good practice. I've added a merge template. --Dafadllyn (talk) 16:55, 27 April 2023 (UTC)

Answered/discussed questions

How do I tag this road?

Hi, I don't know how to tag the road "Francoper Straße" in Hamburg Germany.

The road looks like this: A -> B -> C -> D -> E
A->E
name=Fancoper Straße

A->B
highway=residential max_seed=30km/h lit=yes

B->C
tunnel=yes motorcar=no highway=footway bicycle=public max_seed=30km/h lit=yes

C->D
highway=residential max_seed=30km/h lit=yes

D->E
highway=secondary max_seed=50km/h lit=no

How to insert this to OSM

Possibility 1: 4 Ways

4 Ways (A-B, B-C,C-D,D-E) with properties like above. This means 4 Streets with name=Francoper Straße

I would tag it this way. Some data is duplicated but this is the simplest method and it is clear what is meant and easy for other people to work with. 80n 09:34, 31 Jul 2006 (BST)
This would mean that a motorway is not only one way, but many, because there are many proprerties that change (like max_speed). Sven Anders 14:52, 31 Jul 2006 (BST)
Yes, it is probably best at the moment to start a new way at each motorway junction, otherwise they become too long to be manageable. Also there should be a separate way for each carriageway (both tagged as oneway=yes). 80n 19:15, 31 Jul 2006 (BST)

Possibility 2: Like described above (5 Ways)

One Way (A-E) for the name. The Other ways for the other Properties ....

This is OK, but quite tricky to do and not easy for others to notice or work with. JOSM does not have a good way of showing or selecting from multiple ways. If the client tools provided good support then this would be OK, but I'm not sure how easy it would be for a novice user to understand how things had been marked up. 80n 09:34, 31 Jul 2006 (BST)
This will certainly break many routing-applications that will see 2 ways here.

Nearby roads of same name

Is anyone using a naming convention for same street names in different (nearby) towns? As an example, i have 5 different unrelated streets named "Rheinstrasse" in different towns with less than 5km distance to each other
I see several ways to specify this:

  1. wait for areas to be implemented and create areas for cities
  2. use additional labels like city=
  3. add the city name to the street name (i used this when i split my old tracks, in my archived files i have tracks named as "[Darmstadt] Rheinstraße" or "[Griesheim] Rheinstraße", too track names got lost while uploading/editing tracks)

My personal preference would be (2)

Frank 21:29, 25 Feb 2006 (UTC)
is_in tag can be used, it have been used in many place names, so why not also in streets or other tags that might be similar in different cities --Skippern 23:44, 19 September 2008 (UTC)

Miscellaneous questions

  • Should mini-roundabouts (comprising of just white paint on the road) be drawn as a roundabout?
No. Roundabouts entry has information on how to tag mini-roundabouts Avantman42
  • Where a side road connects to a main road should there be a node with three lines connected to it or should the main road be a continuous unbroken line and the side road just ending at but not connected to a node that is part of the main road?
The side road should be connected to a node on the main road. See the Junctions section above Avantman42
  • If a road can be clearly made out from the aerial photo, but there are also a few GPS tracks, which should be considered to be the more accurate?
The GPS tracks. Landsat, aerial photos, etc, can be some distance out Avantman42
  • Should roads be drawn based solely on the aerial photo if there are no GPS track points?
Yes, but mark them with a source tag Avantman42
  • How should (long) tunnels where no GPS is available be drawn?
In some GPS (for the cars for example), the GPS has one or several gyroscopes so that it is able to know the position. FredB 18:04, 18 Jul 2006 (BST)

Streets with different surface types intersecting?

What is the best way to tag intersections of streets with different surface materials? Are there conventions on how to do that without breaking up the streets at ridiculous level and tagging very short stretches with the crossing street's surface material? Could one e.g. in some way tag the street which keeps its surface type through the intersection? See for example this intersection, where Ystadsgatan's asphalt becomes paving stones while crossing Falkenbergsgatan (which uses paving stones all through its full length). Lyxose

AFAIK there is not yet a method to state which street prevails at intersections (in terms of surface) and which get interrupted. You could do it in detail with area:highway=* and polygons, but not with the linear line model. --Dieterdreist (talk) 17:38, 1 April 2020 (UTC)
For OSM2World, I support surface=* tagged on junction nodes. I don't think it's a standard yet and I'm open to use something else if the community arrives at a different approach, but it seemed like the most straightforward solution. --Tordanik 20:33, 1 April 2020 (UTC)

Adding Wikitravel listings

Hi all, I just wanted to let you know that I'm going to be adding a bunch of hotels, shops, restaurants, and bars in Paris. I'd love to have guidance about how best to go about it, and of course correction if I do something incorrectly. Thanks! -- Mark 15:22, 13 February 2008 (UTC)

Streets with 2 names

I have a situation in my town that the same one road(street) has one name on the left side and the other one on the right side. That has happened because the road divides 2 towns. So how can I handle this? thx Soshial 21:06, 21 June 2010 (UTC)

Tagging Areas

The paragraph on tagging areas says that "For areas adjacent to ways [...] leave a small gap between the area and the way". I think this is wrong.

There are two possible cases: 1.) The area (e.g. a forest) extends right up to the road. In this case the area and the road must share the same points. Otherwise a gap will be visible when the map is rendered, but in reality there is no gap. The rendered gap gives the illusion that there is something else between the road and the forest, but there isn't. 2.) The area and the road do not touch. In this case there must be some other area in between, e.g. a strip of grassland between a road and a forest. If this is the case, the grass must be mapped as a seperate area which shares nodes with the other area on one side and the road on the other side.

If we leave gaps where in reality there are none it is impossible to decide if the gap is real or fake.

I know that it is easier to draw separate nodes. Making things correctly is often more work. But if there is no gap in reality there should be no gap in the OSM data. --R kleineisel 07:43, 4 May 2010 (UTC)

If you were to draw, in addition to the current highway as a way, the surface area of the highway as an area (say, landuse=highway or landuse=road), you would have nodes on both sides of the way currently tagged highway=* and the forest or other area feature would share those nodes. Even if we normally omit that landuse=highway area, doesn't mean that the edge of the feature next to the road would change position. Alv 08:10, 4 May 2010 (UTC)
@Alv: This is an excellent argument against gluing areas to roads. Thanks. To prevent gaps for smaller highways one could fill the gap with the landuse the road belongs to. Say if there is a track between a forest and farmland, the track is used by the farmer. So the track could be within the farmland area and forest and farmland would share nodes. --Lukie80 (talk) 11:03, 20 October 2018 (UTC)

Areas adjacent to ways: source?

At the moment in the section "Tagging Areas" the text says: "For areas adjacent to ways, the consensus is to generally leave a small gap between the area and the way instead of sharing the boundary. The way's nodes represent the center of the way and the neighboring area such as a park does not reach the center of the way."

However, I'm not aware of such a consensus. When and where was this decided? Some other pages, like Tag:landuse=residential and Land use and areas of natural land suggest that both methods may be used. Furthermore I found some recent mailing list posts where this was still discussed. If there are some sources for this rule I'd like to add them as references, otherwise I'd like to remove this part of the text. --Marko Knöbl 14:38, 9 September 2010 (BST)

Don't remove the text. Make a decision one way or the other and document it.

When I started making edits there was no documentation on the issue so I did what made most sense to me and drew parks sharing existing nodes from the ways. Then a senior member (andrewpmk) came and deleted my parks and recreated them with a space in between without any contact or explanation other than the changeset comment "road allowance is not part of park". For example:

http://www.openstreetmap.org/browse/way/39453252/history

I searched through the posts and found in various discussions that the consensus was on andrewpmk's side at the time so I didn't argue. Even though I didn't agree, I fixed up any parks that I had initially made with shared nodes that andrewpmk forgot to delete to have a "road allowance". I then documented the discussion consensus in the wiki so nobody else would have the same experience I had of doing the same work twice. Unfortunately I lost the discussion threads that supported andrewpmk's actions but andrewpmk must surely know about them.

--Fla 2 16:42, 9 September 2010 (BST)

Thank you for your fast answer and for your explanations! As there have been many heated discussions on this issue I think it will be hard to make a decision. I'll contact andrewpmk and ask if he knows if there was any consensus on this issue. If there isn't any I think the contents of this section should be modified until a consensus is reached. --Marko Knöbl 19:05, 10 September 2010 (BST)
I think it depends... If you have landuse areas alongside roads, then there should be a gap, as the road way represents the centreline as your quote says, but the landuse area stops at the edge of the road/pavement/verge/whatever. If you have a landuse area that is adjacent to a coastline though, then it does end at the coastline and you should share nodes/start chopping ways and defining multipolygons/whatever rather than leave a gap. Perhaps consensus is the wrong word, but it is the logically correct way as I see it. --EdLoach 20:39, 10 September 2010 (BST)
Unfortunately andrewpmk has not responded to me. However I think that the other pages I mentioned before and discussions like this one clearly indicate that there is no real consensus yet. Therefore I will change this section of the page to reflect this. If somebody can provide a reference to an "official" discussion where this consensus was reached I will gladly change it back again.--Marko Knöbl 21:13, 23 September 2010 (BST)
If you stick to accurate placement of features, it's obvious where the edge "should" be: if you had a gps with submeter accuracy, and you knew where the lot edge is, and stood there, you would - in many or most countries - be somewhere beside the road. It might even have a hedge or a fence marking the boundary, so place the node there. Where one measurent with a consumer GPS doesn't give enough confidence, we take multiple traces and we can estimate distances and angles between other features. We strive to accurate placement of other roadside things, too, but in many parts we have to, or rather it's more effective to start with a rough approximation. It's therefore not wrong to share the road as the edge between landuses, but it would be wrong to take an already separately drawn landuse edge back to the road nodes. Alv 09:06, 24 September 2010 (BST)
Is there any news on that issue? Gppes 08:14, 08 December 2015 (CET)


I think it is about time that a consensus was layed down on the above issue. The Error-Check routines and Renderers need clarity, as do the mappers. I am sick of Validator and other routines listing possible-please-check-errors. As most of the possible errors are not errors at all I tend to ignore Validator's messages but that is not as it should be.

So please set a standard such as:

    Areas can, but must not, have shared nodes: 
             Logical as the line between the nodes has no width
    Areas and ways (roads/tracks etc.) cannot have shared nodes: 
             Logical as the road/track has a width
    Note: Editing OSM-data becomes more difficult if areas and
          roads/tracks have shared nodes. 
          
          If an area such as forest is adjacent to farmland then the
          editing becomes easier because only one set of nodes needs to be
          geographically moved for two areas. 
          Often is the case that the forests have been entered into OSM
          using aerial imagery and these entries are generally way out
          by up to a few hundred meters. For hikers that is a large
          navigational error.

--Dcp 18:52, 25 September 2010 (BST)


Although the wiki does say "usually" it is too generic. For e.g., landuse=residential it is reasonable to make a gap. It makes editing much easier and you can argue that nobody actually lives right next to the road. If the area is boundary=administrative then it probably should be attached to the highway, because there is another boundary of the other side, and we do not want to make the highway a no mans land. If the area is highway=pedestrian and there are no barriers, then the area should also be attached, so that routers can know that pedestrians can walk directly from the road to the pedestrian area at any point.

I also do not buy the argument that it could be understood that the area extended to middle of the road. If we tag an area with e.g., landuse=forest and put a road in it, then no one would assume that trees are growing in the middle of the road. We usually do not split the forest area in two or make create relations to separate the road from the forest. It is just understood that the highway overrides the forest. It should be the same if there is only forest one side of the road or if the forest is split into e.g., pine on one side of the road and oak on the other side.

Usually the exact width of the road is not known. So if the reality is that the area stops where the highway begin, then we only have one way of tagging that and that is to attach the area to the highway with the understanding that it means that it stops at the highway. Even if the width of the road was known (e.g. from a width tag), it would be practically impossible to map an area that extended precisely to the edge of the highway. I guess we could make a tool for it. But then it would also be a lot of work to adjust areas every time the highways are edited, renders might make curves of the roads smoother than just straight lines, etc. We could also replace all these areas with a new type of polygon-line relations with an outer area extending to the center of the road and the road as an inner member, but no tools would understand that now, and it would gain us nothing because that is how we should understand it anyway (except for boundaries).

In short, we do not map the edges of highways (except for pedestrian). The only way than can work is if roads trumps areas, which is natural for a streetmap. Elgaard (talk) 16:55, 3 November 2018 (UTC)

@Elgaard Your conclusion isn't universally true. Although it isn't all that common yet, we actually do map the edges of highways. See Key:area:highway. Leaving gaps between our 1-dimensional highways (which will be needed for efficient routing for a long time to come) and adjacent areas makes that next step a lot easier. --Hjart (talk) 22:36, 29 November 2018 (UTC)

What counts as dividing a road?

I have many streets in my area where the roadway is one slab of asphalt, but the left and right -hand traffic is separated in some sections by a relatively small concrete "berm". Often this is within a half-mile of major intersections or on high-traffic arterials. These barriers are not islands, but short concrete humps with rounded home-plate (square-bottomed pentagon) shaped cross-section, about 9 inches high and 18 inches wide. It would probably be possible in extreme circumstances for certain high-bias vehicles (SUVs, etc.) to cross these barriers without much trouble, but the intention is that they are never crossed.

Does this qualify as a divided road? Should I split the ways in these places?

We have other streets where, in lieu of the more expensive concrete lumps, they simply paint an 18-inch-wide stripe down the middle. While this is not a physical barrier, it is completely illegal to cross these for any reason e.g. left turns (which is generally legal across the usual double-lines). I'm tempted to think these ought to qualify as divided, because even though they are not physically divided, they are effectively divided by law. But our guideline specifies a physical barrier, so.... it wouldn't seem to qualify. (Well, the dried paint is a physical object, at least!) What to do with these?

TIA - KTyler 09:10, 11 June 2012 (BST)

Storm damaged trails.

I know of some trails that are currently utterly impassable due to storm damage. (Think piles of trees, each 1-2 feet in diameter, stacked maybe 4 or 5 deep.)

It doesn't seem right to just delete the trails: someone might want to use that info to rebuild them. On the other hand, people hiking should know they are impassable, so they shouldn't be left as they are. I haven't been able to find anywhere what to do about these/ what the marking convention should be.

I have no idea whether these trails will indeed be rebuilt: they are not owned by any formal organization, near as I can tell.

Use a lifecycle prefix such as "disused:" or "destroyed:" - if it is a highway=path change it to "disused:highway=path" and add a "note=" . Most of the lifecycle prefixes don't have any special meaning to most renderers or routers - software will just discard them as unknown objects but if the path gets restored it will be much easier to restore it from that state than drawing a completely new one. So you could also invent your own "impassable:" or similar prefix or maybe something similar is already there. Btw you can "sign" your messages on talk pages with "~~~~" which will be expanded to something like - RicoZ (talk) 20:38, 27 November 2018 (UTC)

Two nodes on straight ways, or more?

This page (and Good practice#Keep_straight_ways_straight) recommend mapping straight ways with two nodes at the endpoints of the straight section.

The problem is that routers only sample elevation on nodes (I believe); they cannot interpolate because there's nothing to distinguish straight ways from low-resolution ways. Low-resolution ways could have the lines between nodes falling off the road/path, for example off a cliff with a steep drop in elevation which would throw off elevation estimates. But sampling only nodes is also problematic because it would underestimate elevation if the road is straight but with rolling hills. Granted, one would have to add a node at the peak and valley of every hill for the most accurate estimate, and there's no way to ensure nodes stay there in place...but it will improve elevation estimate accuracy regardless. You could argue this is tagging for the "renderer", but I suppose you could also argue adding only two nodes at the endpoints (to achieve a perfectly straight line) is tagging for the renderer. Roads are rarely perfectly straight anyway.

Not gluing landuse to roads

Would it be helpful to add a paragraph that says: "Please keep landuse and other areas separate from roads, and do not "glue" them to each other." ? The practice of gluing roads and landuse has fallen out of favour with many mappers. --501ghost (talk) 13:03, 2 November 2023 (UTC)

I would support it, but not entirely sure is there is a clear consensus that it is a bad idea (it is a bad idea) Mateusz Konieczny (talk) 15:51, 2 November 2023 (UTC)