logo separator

[mkgmap-dev] Question reg. LocationHook and incomplete data

From WanMil wmgcnfg at web.de on Sat Jan 7 20:36:40 GMT 2012

> Hi,
> the current implementation of LocationHook is somehow unpredictable when
> data in precompiled boundaries is incomplete.
> Assume we have two different elements e1 and e2, and both don't have the
> mkgmap:postcode tag set at the time that currentLevel is set to something
> like mkgmap:admin_level=11. Assume e1 has all other wanted tags, but e2 is
> also still missing e.g. mkgmap:admin_level=4.
> The element e1 will be removed from the quadtree because it is considered to
> be fully worked out, but e2 is kept. Later, e2 may get the value for the
> mkgmap:postcode tag from a boundary which itself doesn't have the value, but
> that "lies in" a boundary which contains a value.
> If e1 was not removed from the quadtree, it might also have gotten the
> mkgmap:postcode tag set to a value from such a boundary.
> I am not sure, maybe I've introduced the problem with one of my patches.
> Anyhow, the question is: What do we want to have? Should e1 get the
> mkgmap:postcode tag , or should e2 not get it?
> Ciao,
> Gerd


postcodes are (or should be ;-) handled in the same way like the 
admin_level tags. So elements are removed only if the mkgmap:postcode is 
set if there is one element left that could set mkgmap:postcode. But I 
remember that this handling was not so straight forward so the chances 
are not very low that this handling has a bug.


More information about the mkgmap-dev mailing list