[OSM-dev] How do I build the latest db schema?

[Frederik Ramm]

Hi,

> Yes, this is the plan.  There is a --write-mysql-change task which is 
> intended to do just that.  I've checked it in but haven't tested it yet 
> so it will probably have a few kinks to iron out.  With the current 
> planet exports, the --derive-change task can compare two planet files 
> and generate a change set.  This can be fed into the 
> --write-mysql-change task which will apply it to an existing database 
> with the previous planet contents and will create new history elements 
> for all affected entities.

Sounds very good. My application will probably be a fresh setup whenever 
I get a planet file, and then incremental updates made by bounding box 
requests. This means I will not really have two planet files to make a 
diff from. What I'll probably need is some kind of "merge" task - take 
this OSM data and if you already have it, ignore it, otherwise overwrite 
what you have. - Detecting deletions will be a problem though: "Take 
this OSM data, assume it is authoritative for the bounding box so-and 
so; if you have data inside this bbox that is not in the new data set, 
then discard it..."

> The replication tasks may take some time to get working.  I'm worried 
> that the main OSM database will have some unusual data that doesn't 
> occur on my local database (eg. nullable fields that I'm relying on).

I'll try to get a deeper understanding of your streaming concept and 
maybe do some work on the merging aspect.

Bye
Frederik