[OSM-talk] Postmortem analysys

[Lester Caine]

Tom Hughes wrote:
> On 09/01/11 10:34, Lester Caine wrote:
>> Nathan Edgars II wrote:
>>>> But why write routers for the one case thats
>>>> > theoretically possible, instead of the millions that are not only
>>>> > possible, but already in existance?
>>> I don't care how the routers are written. I care about people wrecking
>>> the data by merging dupes.
>> And assuming that no nodes at different elevations but the same
>> coordinates are allowed is just crass.
>
> Look people, this really is very simple and I have no idea why this
> thread has managed to go on so long...
>
> OpenStreetMap has, and always has had, a topological model. If two
> physical things are connected in real life then they should be connected
> in OSM by making them share a node. If they don't then that is a bug in
> the data that should be fixed.
>
> If two things which are not physically connected in the real world are
> sharing a node in the database then that is a bug in the data which
> should be fixed.
>
> Routing programs should rely on the topological data that we provide and
> not guess that things which are close are connected.
>
> People merging duplicate nodes should not do so blindly and should check
> what they are doing - in many cases that may mean having to do a
> physical survey or examine aerial imagery to verify the situation on the
> ground.
>
> Unfortunately the duplicate nodes map seems to encourage people to go
> round blindly merging which is why I don't particularly like it. It was
> noticeable that when I was using it and deliberately leaving some near
> me alone because I didn't know the real situation that other people
> would just come round and merge them anyway.

Exactly ... Personally I will STILL use a new node if I need to even if there is 
an existing one with the same XY coordinates UNTIL I can establish if the 
elevation data needs to be different. Many of the comments as to why there is no 
need for the conflict since they can give examples of why the nodes on different 
ways do not need to overlap seem to miss the fact that splitting a way by 
another way such as an admin boundary needs to create a node which may well be 
defined on the admin boundary already, and the new node needs to be an a layer 
below or above that already defined.

The original decision that there should be no duplicate nodes simply ignored 
many of the arguments that there are very good reasons for needing them, then 
tools like the duplicate nodes map ASSUME that the decision takes priority 
rather than allowing 'duplicates' which are distinguished due to their elevation?

I'll be honest ... it's been a while since I had to worry, but I can remember 
having to edit things with the new node moved away from it's real location, and 
then move it back to where it should be. Deleting the other node at those 
bridges would have been equally wrong.

-- 
Lester Caine - G8HFL
-----------------------------
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php