[Openstreetmap-dev] CSV transport encoding scheme

Immanuel Scholz immanuel.scholz at gmx.de
Wed Jan 25 15:27:11 GMT 2006


> Why not just produce two (or more) delimited files packaged up with
> version/generator file in a zip file or some other sort of archive.
> The delimited file format would be based on the database schema
> (http://www.openstreetmap.org/trac/browser/sql/mysql-schema.sql)
> so you'd have a 'nodes' file which would be formated like
> id,latitude,longitude,timestamp,(etc..if you wanted to include more info)
> and 'segments' would just be
> id,node_a,node_b,timestamp,(etc..)

hm... :-| .... :-| .... :-| .... :-)  I like it!

Except, that the structure of the file has to be coupled to the db-scheme.
Stuff, that is in the DB for performance or similar reasons should not be
included in the export data.

As example (as your own suggestion) the "tags" tag.

And choosing the archive type should be done with care. It has to be
something that do not need to know the complete size of every file in
forehead until it starts to encode. It must be able to start encoding from
a stream and spit out the data before the end of the stream is reached
(not buffer everything).
I really want a transport mechanism that can be streamed correctly this

> I hesitate to suggest this, but you'd also probably want something like
> 'nodes_tags'
> id,tag,value
> and
> 'segments_tags'
> id,tag,value

Sounds good too.

> These could then be loaded into any database or spreadsheet for further
> processing.


This instantly became my current favourite idea.

Ciao, Imi.

More information about the dev mailing list