[Talk-ca] [Talk-us] Large-scale Data Imports and Tags on Nodes

Steve Singer ssinger_pg at sympatico.ca
Thu Jan 8 04:00:57 GMT 2009

On Wed, 7 Jan 2009, Sam Vekemans wrote:

> I cc'd to talk-ca list, as it's also relevant.
> So really, all that is needed is the tag that says it's source, and the NID
> tag. ..  Since the python script only really needs to know weather or not
> that point/way actually exists. .. from there, it compares to the OSM tags.
> the tag "nrn:datasetName=Ontario" is that Really needed?

My current feeling is that we don't need to add any attribution tags to the 
nodes we import from GeoBase for the road network since tags everything at 
the way level and we aren't planning in importing any stand alone nodes.

I think we should be tagging the ways with the following meta-data

-The nid the segment came from so people can cross reference it and to allow 
us some options in handling updates
-A source tag attributing it to GeoBase
-Something that indicates the version/release of the geobase data that it 
comes from.  Ie Ontario_4.0 or Alberta_6.0.  We could combine the 
dataset name with the edition  for this.

I'm not sure we need the segment id since the nid is more unique.


> It's only at the GeoBase side that the tag would be useful. ... searching
> for all the roads in the Ontario dataset of the National Road Network. ...
> so the purpose of having this would be;
> 1 - if we wanted to remove all the data (from a planet.osm dump)
> 2 - if we wanted to replace this dataset with a new one, we would have a tag
> to go by.
> the tag "is_in=Ontario" would that not say the same thing?
> So the arguments stated are;
> 1 - filling up the database with extraneous information
> 2 - increases the size of the file download
> That brings us to an excellent point;
> We must be sure to "clearly define" the purpose of adding each extra tag.
> If the tag only serves as a 'cosmetic' tag, it's not needed.
> IMO Tags which get imported should do the following;
> 1 - tell the renderer what the data is and how it relates to existing OSM
> Map Feature standards
> 2 - tell the OSM users WHO imported the data
> 3 - give the python script some cross-referencing tags to 'RoadMatch'
> 4 - be the same type of tags that the OSM user would place in otherwise, if
> the imported data was NOT available.
> IMO So far, i think the tags we have are fine, but it's worth some
> discussions. :)
> Message: 1
> Date: Tue, 6 Jan 2009 11:34:49 -0600
> From: "Ian Dees" <ian.dees at gmail.com>
> Subject: [Talk-us] Large-scale Data Imports and Tags on Nodes
> To: Talk-us <talk-us at openstreetmap.org>
> Message-ID:
>       <3baad6b80901060934x4ab85375o7b2dcdcc39f3c1d7 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> I don't think it makes sense to put tags on nodes for mega-imports. This
> adds tons of extraneous information that fills up the database quite
> quickly. It also increases the size of OSM file downloads by a large margin
> Is it really a good idea to attribute each node? Let's stick with
> attribution on ways. If we're really worried about it, we could do mass
> imports with a specific user and give that user a wiki page to describe
> where the data came from.
> ...discuss...
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://lists.openstreetmap.org/pipermail/talk-us/attachments/20090106/2c2a9aa6/attachment.html

More information about the Talk-ca mailing list