[Geocoding] Local Mirror of OSM Data...

[Anthony]

On Fri, Oct 23, 2009 at 11:34 AM, Tom Hughes <tom at compton.nu> wrote:
> On 23/10/09 16:20, Anthony wrote:
>
>> For my database, I installed rails-port, dumped the database, and
>> edited the dump file to fit my needs (I got rid of a bunch of the
>> foreign key constraints, got rid of the "current" tables, got rid of
>> the "private" tables and columns which I won't have access to, added
>> fields like "isdownloaded" to tables like the changesets table to let
>> me know how up to date the data is, etc.).  I plan to implement the
>> "current" tables initially as a view, and ultimately as a materialized
>> view.
>
> What is your intended use for the data? I assume (given that this is the
> geocoding list) that it's some sort of geocoding?

I have a number of intended uses.  At least one of them involves
geocoding - matching up the parcel information I have from the county
with the highways already in the OSM database.

But my response was meant to be general, as is my cache.

> The reason I ask is simple that the best schema to use for the data is
> generally heavily influenced by what you want to do with it - the rails
> schema is designed to make updating etc easy and is often not a good choice
> for other tasks.

I don't particularly like the rails schema, but I want my cache to
match the original.  I'll create views and/or additional tables to
deal with the other issues.

>> (*) Alternatively, is there an easy way to download *everything*,
>> including current, history, and changeset info, preferably just for a
>> large subset of the world, in my case Florida?  I could probably
>> accept that, but I couldn't find the data.  So, I've basically decided
>> to implement a cache (I call the database osm_cache), and get the info
>> from the API when it's not in the cache (and cache it).
>
> This kind of thing is discouraged - the API is really intended for editors
> and the like not for this sort of bulk usage.

Well, I am an editor.  And I'm not sure if the ways I'm using it
qualify as "bulk usage".  I've been using it to, for instance,
download all the changesets made by *me*.  AFAIK there's no reasonable
way to do that without using the API.

If you have a better suggestion, feel free to suggest it.  Or if you
can point me to a policy which forbids what I'm doing, I'll be happy
to stop contributing to OSM.