[OSM-talk] The biggest violation of OpenStreetMap, ever.

[Christian Quest]

On the first zoom levels, I'm using the capital=* tag to select the country
capitals, then sorting them with decreasing population.
It is a very small number of objects, that can easily be maintained.

The postgis query is here:
https://github.com/cquest/osmfr-cartocss/blob/master/osmfr-cartocss.mml#L2070

If you remove the FR specific part, it looks like:

(SELECT way,
          place,
          name,
          cast(regexp_replace('0' || population,'[^0-9]','','g') AS bigint)
AS pop,
          coalesce(tags->'is_capital', (CASE WHEN coalesce(admin_level,
capital)='2' THEN 'country' WHEN coalesce(admin_level, capital)='4' THEN
'state' ELSE NULL END)) AS is_capital
   FROM planet_osm_point
   WHERE place IS NOT NULL
     AND place IN ('city',
                   'town')
     AND (tags->'is_capital' IN ('country',
                                 'state')
          OR capital IN ('2',
                         '4')
          OR (capital='yes'
              AND admin_level IN ('2',
                                  '4')))
   ORDER BY is_capital,
            place,
            pop DESC) AS placenames


As it is a bit a mess in the capital/is_capital tags so I had to use this
long coalesce/case to deal with different tagging.
As you can see it uses the place=* tag + is_capital/capital + population.
The result looks ok to me:
http://tile.openstreetmap.fr/?zoom=5&lat=43.52781&lon=4.22487&layers=B0000000FFFFFFF
At zoom 4, I just put a black dot for capitals.

Starting at zoom 6 there is an additional placenames layer to "fill" empty
spaces. This avoid large areas with no names at all due to hard cuts in
place=* tags. Many areas in the world have far less population than in
Europe so the stylesheet has to adapt to this.

Compare:
osm.org:
http://tile.openstreetmap.fr/?zoom=6&lat=23.87977&lon=-4.41038&layers=00B00000FFFFFFF
mapquest:
http://tile.openstreetmap.fr/?zoom=6&lat=23.87977&lon=-4.41038&layers=000B0000FFFFFFF
osm-fr:
http://tile.openstreetmap.fr/?zoom=6&lat=23.87977&lon=-4.41038&layers=B0000000FFFFFFF



2014-07-14 11:54 GMT+02:00 moltonel 3x Combo <moltonel at gmail.com>:

> On 14/07/2014, Martin Koppenhoefer <dieterdreist at gmail.com> wrote:
> >> Am 13/lug/2014 um 22:29 schrieb moltonel 3x Combo <moltonel at gmail.com>:
> >>
> >> If osm is missing placename population figures,
> >
> > deducting the importance from population alone doesn't hit it, and adding
> > ranks is generally disputed by many mappers (subjective), so here there
> is
> > no easy solution. IMHO there is not even a complicated solution, as it is
> > indeed subjective how to weight different aspects like economy, politics,
> > transportation, communication, religion, ...
>
> Indeed it's a thorny subject, but delegating the decision to NE sounds
> like a cop-out. Whatever hard data NE used in it decision making
> (population, area, administrative status, connectedness...) should be
> available in OSM, and the subjective algorythm that takes this data to
> output a global place ranking could go either in the style or in a
> common extraction script.
>
> For that matter, how does the osmfr style do its place ranking ?
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>



-- 
Christian Quest - OpenStreetMap France
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20140714/11f7a935/attachment.html>