[OSM-talk] [OSM-talk-fr] Continued aggression against French contributors (cadastre integration)

Christian Quest cquest at openstreetmap.fr
Thu Oct 18 19:52:45 BST 2012


So far, the only explanation about the usesulness of the dedicated
account is linked to tracking imported data or I missed something on
the wiki.

If this is the goal, why small changesets of imported data may not
require a dedicated account ?
This data doesn't need to be tracked ?

I'm also really wondering on the tracking of the data thru dedicated
accounts for such highly split imports (as a reminder, the cadastre
data is split with 36000+ datasets, one for each village/town/city in
France).
On how many accounts will the cadastre data will have to be tracked ?
Is there a list of "cadastre import dedicated accounts" somewhere ?
Maybe a required naming of the account names ?

The more interesting proposal I've seen so far were the "bot/import"
tags. It really brings a benefit for the tracking and makes the
dedicated account requirement outdated.


I also asked some time ago if it was necessary to have a dedicated
account for each source of imported data ?
If tracking is the goal, it seems logical.

If so, I need one dedicated account for:
- cadastre (administrative boundaries + buildings)
- IGN (geodesic points + GEOFLA place=*)
- schools
- RATP (paris public transport, subway station, and we expect 12000+
bus_stops should be available sooner or later)
- SNCF (railway stations and level crossings)
- La Poste (post office locations)
- Nantes metropole adresses (400.000 imported street by street after
crowdsourced reviewing)
and so on... because that's today's situation with more and more
useful datasets to bring into OSM.


Wake up ! opendata is here, now, and the more (useful) datasets we
find the more it is clear that a mass import is not possible.
We've learned from CLC and Tiger imports... and the map is not blank
as it used to be.


Here is my proposal... have separate guidelines for mass imports and
for split/shared/crowdsource imports.

When a full dataset is imported more or less "as is" by a very small
number of contributors (possibly just one) in such case, YES, a
dedicated account is a real benefit. With ONE dedicated account, you
track all the imported data.

When the dataset needs to be reworked manually, integrated sometime by
reviewing each object one by one, or sometime group by group, where
the works is share by a lot of contributors, in such cases frankly I
don't see the benefits of the dedicated account(s). The bot/import tag
on changesets is much more efficient.

Christian



More information about the talk mailing list