[osmosis-dev] nodes column in way table and way_nodes table

Michel Seuthe m.seuthe at googlemail.com
Wed Dec 21 13:31:53 GMT 2011


2011/12/21 Igor Podolskiy <igor.podolskiy at vwi-stuttgart.de>

> Michel,
>
> please reply to the list, _not_ just to me in person. Thank you.
>
>
> On 21.12.2011 14:00, Michel Seuthe wrote:
>
>> Hmm okay, but that means all references from the way to it's nodes are
>> stored in way_nodes and in way table. Which would be a mass of redundant
>> data...
>>
> Why do you think it is redundant? It isn't. The node sequences of the ways
> aren't stored anywhere else in the database. The snapshot schema is as 2nd
> normal form as it can get...
>
> The ways table store only the assembled line strings, i.e. the _geometry_
> of the way derived from the node sequences. If you don't need this or if
> this is a problem for you, use another schema. APIDB, for example, does not
> store assembled geometries.
>
> Bye
> Igor
>

The "blank" snapshot schema doesn't store any assembled geometry data. Just
if I add the linestring-column with the provided script-file
(pgsnapshot_schema_0.6_linestring.sql). But it sure does store a
bigint-array typed column in the *way *table and references the ways to
it's nodes in the *way_nodes* table. And the bigint array itself has a
sequence (it's index). So the* way_nodes* table is fully redundant to the *way
*table with it's columns *id *and *nodes*.
This is a bit confusing, but i hope you understand. I'm confused by this at
all.

Cheers.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmosis-dev/attachments/20111221/e9cea4ba/attachment.html>


More information about the osmosis-dev mailing list