[OSM-talk] calling all python coders for slippy map help

[Schuyler Erle]

* On 16-Nov-2006 at 12:08AM PST, To Martin Daly said:
> * On 16-Nov-2006 at 12:01AM PST, Martin Daly said:
> > > Sounds like planet.osm isn't, er, well-formed. Any ideas?
> > 
> > Try Jonas' UTF8 sanitized version, here: http://www.mozoft.com/OSM/
> 
> I'm fetching it now; will try and let you all know. If someone else
> already has the data and is willing to give it a whirl, it'd be much
> appreciated.

The punchline is that my version of the script ran for 5 hours on a
machine of reasonable capacity, then died on a missing node. It's
possible that enabling the BDB cache would fix this (or making the LRU
cache bigger), but in that time the script only spit out 42,000 ways.
If my understanding of the database size is correct, that suggests
that the conversion run using this method would take three weeks.

It seems very probable that the big slowdown is the LRU cache's
maintenance of its internal heap.  Amaury made noises on IRC about
throwing data away more intelligently.  Perhaps he can expand on this?
I don't fully understand the format of the planet.osm file.

SDE