Let’s say I have some fields adjoining an area of woodland. Along the boundary is an intermittent stone wall or a stone wall which then becomes a fence.
I’d like to show the two areas and also where the walls and fences are.
I draw a bound area representing the woodland (I can use landuse=forest or natural=wood, the difference is that they will render differently in Mapnik). I can draw a separate bound area for the fields and can tag it with, say, landuse=farm. I can nudge the bounding line of the field boundary close but not overlapping the bounding line of the woodland. In between the slight gap between the two is where I can draw the stone wall or the fence. I end up with three lines squeezed tight representing the boundary I’m trying to describe.
Option 2. I can use a common line between the two bound areas (the two areas sharing nodes along the extent of their commonality). More efficient yes, but how can I represent the wall? Or the fence?
Why represent the wall? When walking, the pattern of walls and field boundaries is one of the best local indicators of where you are.
I’m not too worried about differentiating between fence and wall - neither can be crossed. Anyone following a path would obviously aim for a gap or a stile or a gate or something similar which can be tagged.
I’ve never bothered tagging an area as a field, or Landuse:Farm, and the thought of doing so makes me imagine stupidly complex bits of map data that are very hard to work with, just to produce quite a useless bit on the map. I don’t think the tiny distance between the wood and the wall/fence matters particually either, and since it marks the woods edge I would add it to that way, or make a 2nd way from the same nodes.
When a wood has a fence around it though I do usually tag this, since it’s handy to know. An enclosure is different from open woodland when it comes to walking in the straightest line from A to B.
I currently have all hedges/fences/walls/gates/wooden-fences/etc.etc.etc under one key. =yes tags are really messy. The =yes is a pointless element, and the key (which should be used to group similar things), is replaced by a specific thing. The =yes tags are becoming concerning common. The key I’m using is ‘border=’ which is just a temporary word until there is a standard word. Perimeter/boundary would also do. The wording doesn’t really matter to much, the tagging structure does. Differentiating between walls/hedges is very important, because if anything on a map is too common it becomes worthless to a lost person.
The key/value arrangement in OSM should be used to say something like ‘this ‘BUILDING’ is a ‘GARAGE’’ but instead these yes tags are saying ‘GARAGE?..oh YES…it’s one of those’. And it also presumes no multiple definition for the key word.
My inspration is the 1:25000 OS First Series maps where there are contours and wall boundaries. The thin grey lines can equally be fences but these OS maps date from an era when barbed wire fences were few and far between on the moors and open spaces. The 2nd series adds some good features too but sadly all discontinued…
A single line is all that is needed for walls and the walker when checking the map against the real thing would make the connection: these are dry-stone walls - here is the juction of these walls - so this is where I am and this is the route I want to take. The problem is that not all walls are boundaries (though once they probably were - or else why have them?)
Currently there is nothing on OSM that renders just walls. The available tags seem to force me into landuse=farm which I don’t care much for, or natural=wood or natural=heath - all of which don’t hang together well.
OSM (implicit in its name) clearly derives from street mapping - and I guess what i’m trying to do is create localised maps that are for the walker the equivalent of the cyclist in gravitystorm.
I agree the “xxxx=yes” tag is not only messy but also bad structurally. Yet I find myself having to use it (as in building=yes) in order for anything to render now rather than in some distant future (or never?).
Hopefully the incremental improvements and incremental convergence towards a unified set of tags will not only make mapping more easy (especially for newbies like myself) but create good-looking maps that are consistent worldwide. But that’s probably a long way off.
Hi and yes, feature creep is inevitable. I once developed a complete database system for a client and had to get quite tough about feature creep. It was so easy for the client to want this, then that, then something else… By the end of it I’d become a slave to “if only…”.
However, open source will always have as many diverse directions as there are subscribers and there is nothing preventing anyone wanting more. gravitystorm has obviously linked into the OSM dB and added contours and the like and successfully too. An equivalent for walkers and hikers would be great too. We all have different end-games I suppose.
The source is OSM and so long as this remains accurate and representative then it can find a 1001 end uses. The debate as to the “best” tags will no doubt continue but, over time, there will be convergence and an accepted set of standards will emerge as representing best practice. This alone will be so helpful to newcomers like myself.
With border=cliff and border=fence I have a problem. And it is a simple one: what I see is a fence or a wall or some other physical feature. The feature may or may not coincide with a border which in turn may or may not coincide with an international, national, country, county/state, parish, or any other “border” or “boundary”. I see and can touch a wire fence - but the border is a construct - a different entity altogether.
I think there is a place for borders but this is a different “layer” or at least a different overlay on top of the purely physical. The OS maps of GB - especially the First Series and parts of the 2nd series remain an inspiration. Relatively simple and most elegantly executed. (Well it should be - given the taxpayers’ inputs into Ordnance Survey - but I need to climb down from that perticular soapbox).
Maybe too many people are trying to make OSM too many things to to many people?
But’s that understandable when the underlying product is soo good!