[OSM-talk] Residential areas

[Richard Fairhurst]

Quoting Tom Chance <tom at acrewoods.net>:

> Being a programming dunce, how hard would it be for a tool to know if a
> feature is within or adjacent to a landuse area?
>
> The answer, to my mind, partly determines what we do with abutters.

I've said this before ad infinitum and apologise for repeating myself,  
but: in the medium/longer term, I do not expect many tools to work  
with "unprocessed" OSM data - just as they wouldn't with the full OS  
MasterMap database.

Rather, OSM will be the base dataset from which people can extract,  
process and simplify their own dataset. These datasets may well  
contain an equivalent of abutters, derived from the land-use data,  
_if_ it's appropriate to their needs (e.g. routeplanner: "you are now  
entering the city of Wolverhampton"). Real-world tools will use these  
datasets, not the full planet.osm.

Can a post-processing script determine whether or not a road is within  
a landuse area? Yes, unquestionably. Is it computationally intensive?  
Yes, it is. But you only have to go through that process when creating  
your simplified, derived dataset. The tool itself works on the derived  
dataset, so doesn't need to carry out the "which land-use area am I  
in?" calculation every single time.

(All of which means: land-use represents the situation on the ground;  
abutters is a compromise whose limitations are becoming apparent;  
let's go for land-use.)

FWIW, I think this whole approach is why Google Maps is so successful.  
Google has gone through a whole load of processing effort to turn the  
raw TeleAtlas data into pre-rendered tiles that (a) look pretty and  
(b) can be served instantly.

cheers
Richard