[Geocoding] Local Mirror of OSM Data...

[Tom Hughes]

On 23/10/09 18:58, Anthony wrote:

> You're likely already using that disk space, though.  If you used a
> materialized view for the current_nodes table, you wouldn't need the
> current_nodes table.  You're basically already implementing
> materialized views, you're just doing so in the code rather than in
> the database, which if nothing else makes it more prone to mistakes.

Oh I absolutely agree that the current tables are not ideal - they 
predate my involvement with things so are not my choice. My long term 
aim would definitely be to replace them with views.

There isn't even any real reason why it would even need to be 
materialised - at most partitioning the table on live/historic and/or 
having conditional indexes should be enough.

Tom

-- 
Tom Hughes (tom at compton.nu)
http://www.compton.nu/