[OSM-talk] Keeping imported data updated with source changes

michael spreng talk at osm.datendelphin.net
Sun Jan 11 12:11:44 UTC 2015


Hi Wiktor

I maintain an import of public transport stops in Switzerland. These
public transport stops have a unique identifier (uic_ref) which is used
for example in time table applications, so we include it. But I spend
quite a bit of time maintaining these refs, as mappers usually don't
know and care about them.
I understand your reasoning but I would advise against introducing this
reference in your Import. It doesn't feel right in the OSM database and
it will cause quite a bit of work to keep them tidy.

That being said maybe you shouldn't aim at a 100% solution. It is very
kind of you to want to make it perfect, but I think a solution which is
a little bit simpler but which misses a few corner cases will be all
right. (Making the corner cases miss some data to import, not destroy
OSM data)

I'm biased, because I maintain the tool, but: I think it helps the power
mappers to have a map where they can compare OSM data with the other
datasource (see http://didok.osm.ch). It shows a few things: it shows
stops which are only in OSM, and in a different colour stops which are
only in the data source. And it shows a line between OSM data and the
data from the other source, making it easily visible where something is
wrong. It also has a separate database where one can mark stops as
invalid, something you will probably need as well.

And then, from time to time, I propose some specific mechanical edits
fixing things that are often wrong in OSM. Of course this leaves a lot
of manual work, and I'm lagging behind considerably.

I hope that helps, I wish you a lot of success with maintaining Polands
addresses.
Michael



More information about the talk mailing list