[Tile-serving] [osm2pgsql] Reduce memory usage of node cache (#360)

[Paul Norman]

I'm not sure we need to wait for what I'm describing to merge, but here are my plans. For valid combinations of slim, flat nodes, cache strategy, and `--cache` being zero or not, do the processing section of an import with the null backend. Then use the C++ api to fetch positions from the middle.

The test file would have a full dense block, a partial dense block, and some sparse nodes.

Out of order and negative nodes would be nice to test, but as the correct behaviour with some combinations of options (`--flat-nodes --slim --cache 0`) is to throw an exception, this needs to be its own file.

---
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/pull/360#issuecomment-112990863
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20150617/2b1b57e3/attachment.html>