Frédéric Rodrigo

Here we have one real case https://github.com/osm-fr/osmose-backend/issues/1735 Maybe we can use the output of backend/osmose_config.py and compare with data base content (but there is external analysers). @jocelynj better idea?

> I see we have a "timestamp" column on table "class" on frontend. Could it be used to remove old classes? Right. But it solves only one part of the...

Note 1. We already have a validator. This should not be for detecting issue, but only for valid MP. Note 2. I think `ST_MakePolygon` done most of the work

> Regarding note 2: correct for the simple cases, but unfortunately it doesn't seem to support multiple outers or composite (non-closed) linestrings forming a closed loop. (And I personally have...

I try an implementation. The naive query is, but it work only with valid data in OSM ```sql CREATE TEMP TABLE multipolygon AS SELECT id, ST_BuildArea(linestrings) AS poly FROM (...

> > It is not so fast. But with that I am able to build 183 716 MP over 195 072 of France. Mote than 10 min. > > So...

I made a more shorter and false safe version. ```sql CREATE OR REPLACE FUNCTION build_multipolygons() RETURNS VOID AS $$ DECLARE mp RECORD; BEGIN DROP TABLE IF EXISTS multipolygons; CREATE TEMP...

> After polygon_small, probably also the following other analysers can benefit > And maybe more. The common table `buildings` could also be improved.

@Marc-marc-marc can you explain what's wrong? What need to be changed? We may add an GeoJSON Maproulette compatible export on Osmose side.

@Famlam the issues are gone, please can you re-check?