[Tile-serving] [osm2pgsql] Conversion to C++ (#187)

[Paul Norman]

On machines with *very* high IO concurrency, I'm seeing slightly longer `ORDER BY`/index times

![image](https://cloud.githubusercontent.com/assets/1190866/4490057/1f43cd2a-4a25-11e4-89bb-c7ac6c70fddc.png)
i7-4770, RAID0 array of 2 SSDs

As this only shows up on machines with RAID arrays of SSDs and we want to rewrite this part anyways to avoid `ORDER BY` and do something sane, I'm not too worried.

The effort reducing the problem to a cause is better spent switching from `ORDER BY` to `CLUSTER` (#99), disabling `ORDER BY way` "clustering" completely (#102) and going by `ST_GeoHash(ST_Transform(ST_Envelope(way),4326),N)` (#87).

I would rather get the C++ merge in and do a separate PR for this, given the huge complexity we have already.

---
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/pull/187#issuecomment-57613985
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20141002/b74554c8/attachment.html>