[Talk-ca] How should a noob proceed?

Steve Singer ssinger_pg at sympatico.ca
Mon Jan 31 02:58:42 GMT 2011


On Sun, 30 Jan 2011, Samuel Longiaru wrote:

> 
> then realized that doing so for each import would eventually lead to massive relations of
> wooded area.  So would you agree that I should leave the bounding ways alone, only working on
> merging or joining cross-boundary objects such as roads, railroads, etc., or should I treat the
> bounding ways somehow ?

a) I would merge the nodes between on the objects so they are 'connected'.

For example, I think your nodes 1127393723 and 1127420928 probably should be 
merged.  If you zoom in with JOSM very close to those nodes you see that 
there is a gap between the north and south halves of the lake.

To merge a node in JOSM, zoom out enough so the mouse covers both nodes. 
Then "left click" to select a node, then use the centre mouse button(click 
it) Then hold down on "CTRL" and you  can now mouse over the other node (a 
popup should have appeared listing all the objects near the mouse) and 
select it with the left mouse button.  Both of the nodes should now be 
selected.  If your press "M" it should merge the two nodes.

Doing this might fix a lot of the issues your seeing with the validator.



For something like a small lake where the lake is going to span 4 tiles but 
the lake isn't that big I'd merge it.  But a wooded area that when merged 
takes spans 100km with hundreds of nodes should probably be split at some 
point.

(always merge nodes that represent the same point in space, but the 
associated ways don't need to be merged).


> 
> 2) I am surprised in that the registry between the different objects in the CanVec data is not
> better than it is... at least in this area.  For the most part, the boundary of the wood
> (whether outer or inner) seems to be shifted westward in relation to the lake outlines which
> are at least fairly consistent with the Bing imagery.  This internal shift within the CanVec
> data seems to triggering a lot of errors... overlapping areas, crossing ways, etc.   The shift
> seems to line up across both sheets.  Is this expected, and if so, should I just correct errors
> as usual, using the Bing imagery as a guide?  Or is this error something that happened when
> converting the data to OSM format?
>

I've seen this in other areas as well but I don't know what the cause of the 
shift is.

In many places the Bing imagery is better resolution and newer than what was 
used in Canvec.  If you think something from Canvec can be improved/fixed 
based on the Bing imagery then you should feel free to do so.  However 
sometimes the Bing imagery could also be  slightly mis-aligned (If you 
determine this you can realign the imagery in JOSM for your editing 
session).   If Canvec, Bing and your GPS traces all show something to be in 
the exact same spot then you can be confidient that the object is positioned 
right.  If all three sources show a slightly different position then it is 
hard to know which (if any) are correct.


> Sam
>


More information about the Talk-ca mailing list