[OSM-dev] SOTM EU Imposm presentation: osm2pgsql memory usage with large extracts

[Oliver Tonnhofer]

Hi Paul,

thanks for your clarifications. I did indeed only explained how the DENSE allocator works, because that was the way it worked back then. I now think I failed to make that clear. I should have focused more on how Imposm works instead of reasoning why we wrote it.

On 18.06.2014, at 18:44, Paul Norman wrote:
[…]
> While re-checking the presentation for memory usage, I noted a mistake about flat-nodes. Flat nodes is not like normal mode, it is part of slim mode, but with the nodes table moved from PostgreSQL to on disk.

I was referring to how it allocated/writes the nodes: as a 'simple list' instead of a database.

Cheers,
Oliver