[Tile-serving] [osm2pgsql] Make more of osm2pgsql threading aware (#110)

[Paul Norman]

So, I ran diffs on a full planet. Unfortunately the results with the normal branch 5ba712d15ba611e502840374b1bacbc72d830042 and threading a5755430392af3459d06ede91e9498b10bd18ae1 are not the same.

They're pretty close. `planet_osm_ways`, `planet_osm_rels`, `planet_osm_point`, and `planet_osm_polygon` all pass my tests, with all the statistics I compute being the same with both branches. `planet_osm_line` and `planet_osm_roads` differ. If forced to guess, I would guess it's something related to splitting long ways.

The relevant queries are
```sql
SELECT COUNT(*), SUM(ST_Length(way)) AS st_length FROM planet_osm_line;
SELECT COUNT(*), SUM(ST_Length(way)) AS st_length FROM planet_osm_roads;

SELECT COUNT(*), SUM(ST_Length(way)) AS st_length FROM planet_osm_line WHERE highway='primary';
SELECT COUNT(*), SUM(ST_Length(way)) AS st_length FROM planet_osm_roads WHERE highway='primary';
```

The results are
```
Test                  Normal N=8 604      Threaded N=8 604   Difference
line count            82988137            82998741           -10604
line length           115470181512.839    115585071070.389   -114889557.6
road count            7743627             7744565            -938
road length           41246103355.6413    41261720915.1192   -15617559.48
primary line count    1320509             1320575            -66
primary line legnth   2734979047.75017    2734985654.309     -6606.55883
primary road count    1320509             1320575            -66
primary road length   2734979047.75017    2734985654.309     -6606.55883
```

---
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/pull/110#issuecomment-32048287
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20140110/053d7e36/attachment.html>