Steve Bennett wrote:
Anyway, before I take a sledgehammer to map_features.xml
Thanks for volunteering for this one - it's definitely something that
needs sorting now that Bing imagery makes it more likely that people
will map areas rather than just POIs. There's no need to take a
sledgehammer to it, just (I'm sure you know this) make sure the
particular feature has both point and area elements, thus:
<feature name="Marina">
<category>water</category>
<area/>
<point/>
<tag k="leisure" v="marina"/>
<inputSet ref="names"/>
<inputSet ref="common"/>
</feature>
1) Is there a particular philosophy regarding POIs vs other features
that I should be aware of?
Not AFAIK.
2) Are there reasons to group all POI features together, rather than
putting the POI and area-version of each feature together (or, where
possible*, merging them)?
Definitely use a single <feature> element with both <area/> and <point/>
subelements, rather than using two separate <feature>s.
If the tags are divergent, then the syntax to use is:
<point>
<tag k="amenity" v="frog_bothering_ground"/>
</point>
<area>
<tag k="amenity" v="frog_bothering_ground"/>
<tag k="frogs" v="yes"/>
</area>
This isn't supported yet (see line 929 of map_features.xml) but it
should be. :)
3) Should we take into account the likelihood of any given POI being
mapped as a node (ie, supermarket is far more likely than vending
machine), or take a blanket approach that all POIs can be areas?
Don't assume that all POIs can be areas. <point/><area/> is
self-explanatory: overloading <point/> to mean "oh, it can be an area as
well" would be counter-intuitive.
cheers
Richard
_______________________________________________
Potlatch-dev mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/potlatch-dev
