[OSM-dev] (Partially) duplicate ways

[Joachim Zobel]

Hi.

I have searched for ways that are contained in other ways (see attached
SQL) and found 10000+ (5000+ of length > 2) in my local database
containing a post 0.6-migration map (from 2009-04) of germany. It seems
there are quite a lot of obvious inconsistencies like e.g. the ways
33816455, 33816484, 33817385, 33817396 or the ways  33309970, 33309994,
33310274, 33310413 or 33598434, 33598740.

The latest example (ways 33857470, 33857473) is from 2009-04-30. It
looks like this is not a pre-0.6 problem.

I did find the first example because they behave weird when split with
JOSM, so they are obviously harmful. 

Is anybody else currently working on this topic? 

Sincerely,
Joachim

        -- Find ways contained in other ways
        -- Attention: takes hours, even on a map of germany
    SELECT wnl.id, wnr.id, COUNT(1) AS cnt
      FROM current_way_nodes wnl
      JOIN current_way_nodes wnr
        ON wnl.node_id = wnr.node_id 
       AND wnl.id < wnr.id
  GROUP BY wnl.id, wnr.id
    HAVING COUNT(DISTINCT (wnr.sequence_id - wnl.sequence_id)) = 1
       AND COUNT(1) IN (SELECT COUNT(1)
                          FROM current_way_nodes
                         WHERE id IN (wnl.id, wnr.id)
                      GROUP BY id)
    -- AND COUNT(1) > 2