[OSM-talk] I've removed segments from the map features wiki page

John McKerrell john at mckerrell.net
Mon Jul 9 15:21:07 BST 2007


On 9 Jul 2007, at 15:00, Matthias Julius wrote:

> Nicola Ranaldo <ranaldo at unina.it> writes:
>
>> On Monday 09 July 2007 14:58:49 Tom Chance wrote:
>>> Ahoy,
>>>
>>> On Mon, 9 Jul 2007 14:39:48 +0200, Nicola Ranaldo  
>>> <ranaldo at unina.it> wrote:
>>>> Sorry if i repeat an old question.
>>>> If i have a way with the last few segments tagged as  
>>>> one_way=yes, how may
>>>> i reproduce this scenario without segments tagging?
>>>
>>> Split the way into two smaller ways, and tag the one way part
>>> appropriately. JOSM has a "split way" function, download the  
>>> latest version
>>> if you can't see it.
>>
>> Ok, but how renderers, editors and other softwares may recombine  
>> them?
>
> Unfortunately, they don't.  That is the problem that superways or
> advanced relationships are supposed to solve.
>
Mapnik does, osmarender isn't intelligent enough to do so yet (or not  
in the T at H version).

>
>> It could be difficult to add a tag to the entire set of ways if
>> software does not have a topological reference to them.  It could be
>> a problem to show names and refs too or to evaluate the lenght of a
>> way and so on.  Just some considerations.
>
> This is not trivial.  There could be any number of parts of that
> highway that are partially overlapping that require different
> tagging. Like you have a certain speed limit, then the number of lanes
> change, then there is a bridge, then another speed limit, then it
> becomes oneway, then there is a tunnel in the middle of which there is
> an exit and the number of lanes change again...

Ignoring the speed limit and lanes bits there because they don't  
(currently) get rendered, what you've just described would mean 3  
changes in the rendering of that road. As far as the renderer is  
concerned it would be 3 different things, so I don't really see why  
it needs to be grouped by a "superway" (which of course would be yet  
another thing we'd need to add to the data, not exactly reducing  
complexity).

John




More information about the talk mailing list