[OSM-dev] osm2pgsql for 64-bit IDs

Frederik Ramm frederik at remote.org
Tue May 24 16:25:10 BST 2011


On 05/24/11 17:05, Igor Brejc wrote:
> Not necessarily the same thing. Writing protobuf reader is still much
> easier than implementing something like
> http://www.sqlite.org/fileformat.html and
> http://www.sqlite.org/fileformat2.html

If you say so - I thought that dealing with protobuf files without using 
any of Google's libraries was also a challenge.

> I'm not talking about OSM-specific data transport format, just something
> that would:


I still fail to understand what the problem is. If the problem is OSM 
specific data transport, we have solutions for that. If the problem is 
"general data transport for geodata", then maybe this is out of scope 
for this mailing list, and anyway one would first have to make thorough 
analysis about what requirements there are out there, outside of the OSM 
world, for data transport for geodata. And if the problem is 
"accessible, somewhat indexed storage of geodata for processing and 
display" - what shapefiles currently are - then I believe that 
spatiallite and ESRI personal geodatabase are probably the future.

> So I presume Geofabrik will switch to providing spatialite extracts
> instead of shapefiles once the 64bit switch becomes a reality?

Unsure. But OSM IDs are not needed in shape files so we could drop them 
or simply write them as a text field, and as for number of records - we 
don't do shapefiles that large anyway.


More information about the dev mailing list