[Tagging] [Imports] New module to merge-sort imports over time (osmfetch python)

Vincent Pottier vpottier at gmail.com
Fri Aug 26 22:04:23 BST 2011

Le 26/08/2011 04:48, Bryce Nesbitt a écrit :
> On 08/25/2011 05:39 PM, Martin Koppenhoefer wrote:
>> if the data is "mastered elsewhere", don't import it into OSM.
> If OSM is just a unique repository of hand collected data, sure, 
> that's right.
> If OSM is an end result (e.g. published maps) then we should be 
> talking about how valuable the particular data is to viewers... or 
> creating a way to merge the hand-collected repository with other 
> quality repositories to produce useful mapping products.
> Which is the more important goal?  I hope the script I'm working is 
> compatible with either goal, and helps by improving the quality of 
> imports that are done.
>> even if you can't be convinced to put this information in the
>> changeset (which would be preferable as Frederik told you as well),
>> please consider using just one tag. E.g. source:license is not
>> important once the data is in OSM: all OSM data is distributed under
>> the same license. Source:website is information that doesn't belong to
>> the dataset IMHO, it would be better put (together with the license
>> information) on a wiki page.
> source:license was meant to point to the license documentation that 
> permits the import.  I don't mind leaving it out, or moving it to the 
> changeset.
> source:pkey has to remain on the node/way, as that runs the conflation 
> process.
> I will extend JOSM as needed so the changeset tags can be set 
> automatically (the whole goal here is automation and repeatability).
If I understand well, the set of data (car sharing location) first 
imported is already a conflation of existing data in OSM and new data 
given by the agency.
Some people said that the source tag seems more accurate on the changeset.

So maybe the source=my_agency may be wrong when the data is coming from 
a previous conflation, but I need it for maintenance.
If I still understand, the function of the tag on the object (way, 
node...) is to maintain synchronisation, and when the tag is destroyed, 
the conflation is stopped.

So why not rather use a sync:*=* tag ?

The sync key is still not used (on taginfo Data from: 2011-08-26 00:00 
UTC). Write a proposal... I would approve.

More information about the Tagging mailing list