[OSM-dev] [OSM-talk] initializing an apidb database

[Paul Norman]

On 9/8/2014 10:50 AM, Richard Welty wrote:
> i have the rails port installed (with postgresql) on an ubuntu 14.04
> instance in the EC2 cloud. i have an extract (made of state boundaries
> taken from TIGER 2013 data using ogr2osm) which i installed in
> the db using osmosis. since osmosis insisted on positive ids i flipped
> the sign on the extract before installing (using -read-xml and -write-apidb)
>
> i can use curl to fetch boundaries from the db using 0.6 api calls. but
> when i point a copy of JOSM at it and try to download, i get a message
> about an invalid changeset number of 0. what am i overlooking?
> obviously the db isn't quite right.

cc'ing dev@ since this is more of a dev issues than a talk issue.

ogr2osm has some hidden shoot-self-in-foot options for this situation, 
--positive-id, --add-version, and --add-timestamp. These are hidden 
because they are quite specialized, and can seriously screw up live OSM 
data if misused.

For the changeset, I'd take a PR to ogr2osm to add a --changeset option 
that allows you to specify a changeset. It should be basically a 
copy/paste of --add-version.

Otherwise, some SQL could change the changeset number to 1.

If you're just doing state boundaries, the easiest way might be to 
create an account on the instance and upload to it with conventional tools.