[Tile-serving] [osm2pgsql] Don't let invalid geometries into the database (#428)

[Sarah Hoffmann]

My main point was that this is not a bug fix. You like to change established behaviour. That means this is certainly not a backport candidate. You should only backport bug fixes.

The interesting thing about st_buffer(0) is that it creates nowadays mostly empty polygons when it can't fix the geometry (e.g. http://www.openstreetmap.org/way/165367703). That's why you see so few invalid ones. I'd like to see that addresses jointly but I can live with just this change.

Third point: st_isvalid is known to be slow, I don't know about geos' is_valid() function. Did you see any performance impact because of the additional valid checks?

---
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/pull/428#issuecomment-129487595
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20150810/d294923b/attachment.html>