Hello Carles, There is no definite way regarding the use of source:pkey and source in general but I believe that it is better to add the source to the transaction in this case. Source could be added to the POI in itself. Regarding automated push to OpenStreetMap, I would make sure that you have a mechanism to roll back specific POI if required because you are always at risk of having someone entering potentially bad data in OSM. I would recommend discussing with the DWG for better advice. I believe your logic should be revisited as I don't see why you would need to store your id in OSM in the first place. I would probably use the following flow instead: a) We receive a POI from the user and then we add in our database (so it's available to ourselves) b) We regularly push the POIs to OSM (using the API) c) When the API returns the id of the POI pushed, I update my database with OSM id d) When we update OSM (once a week, automatic process): we should remove from our DBs the POI that already exists in OSM
Effectively roughly the same logic that you suggested except you don't export your own id and rely on OSM ids. Please be aware that relying on ids can be difficult in OSM in the first place as you are never sure that they may not change (Someone can delete it and recreate it). This is also true for your own ids if you import in OSM (someone can easily remove it and they will once they will realize it is an external id). I would argue that using OSM id would make your app slightly more resistant in the first place. By the way, this mailing list is very very low volume (probably the first email in years that I have seen). Emilie Laffray On Wed, Dec 3, 2014 at 3:22 PM, Carles Pina i Estany <car...@pina.cat> wrote: > > Hello, > > TL;DR: I'd like to use source:pkey=xxx and source=app_name to remove > local POIs that I create to OSM. Is this the correct way to do it? Or > should I do something else? (e.g. fuzzy matching by location :-( ) > > Longer explanation: > > If this is not the the right list to ask this question please let me > know - and I'll go to the correct place, apologies. I write to the > London one because is where I live. > > Friend of mine and I are preparing an app that will use OSM map and > data. We want to contribute back to OSM and I have a question regarding > the workflow. > > The app will show some POIs and at least a few people will be > able to add POIs that doesn't exist at the moment (let's assume that > only trusted people will be able to add POIs, people that know how OSM > works and the consequencies of adding POIs). > > I've been thinking of doing something like: > a) we receive a POI from the user and then we add in our database (so > it's available to ourselves) > b) regularly we push the POIs to OSM (using the API) > c) when we update OSM (once a week, automatic process): we should remove > from our DB the POIs that already exists in OSM. > > I thought to tag the POIs that we create with source:pkey=$ID and > source=name_of_app (this is similar to what it says on > http://wiki.openstreetmap.org/wiki/Osmsync webpage). > > My question: what's the policy of using source:pkey=$ID, source=app_name > ? Can be used this way or not recomended to be used this way. > > Thank you, > > -- > Carles Pina i Estany > Web: http://pinux.info || Blog: http://pintant.cat > GPG Key 0x8CD5C157 > > _______________________________________________ > Talk-gb-london mailing list > Talk-gb-london@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk-gb-london >
_______________________________________________ Talk-gb-london mailing list Talk-gb-london@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-gb-london
