[OSM-dev] Source tag on each node

[Jochen Topf]

On Sun, Dec 19, 2010 at 09:59:36AM +0100, Wyo wrote:
> I've just realised that rather many nodes have their own source tag,
> even if they are just part of a way with identical tag. So I guess a
> rather large part of the data consists of these source tags. While
> this isn't a problem for files (thanks to compression) it enlarges
> the database quite a bit.
> 
> Are these tags really necessary or could they be remove as long as
> the way has a source tag? Could anybody provide an estimate how much
> space reduction could be gained? Could anybody create a report on
> how many nodes have identical tags as their way?

As you say tagging source on nodes is generally not necessary if the only
thing they do is support a way that has a source tag.

http://taginfo.openstreetmap.de/keys/source#values and
http://taginfo.openstreetmap.de/sources/db tell you some numbers: There are 34
Mio source tags on nodes. Note that the percentage for the node is given
relative to the number of nodes that have at least a single tag. So about 50%
of the nodes with at least one tag have a source tag. Only about 8% of nodes
have any tag. So that means only about 4% of all nodes have a source tag.
The 34 Mio source tags are about 6% of all tags (527 Mio) in the database.

So in the very "best" case we would save 6% of tag storage space, but there
is much more stuff in the database and of course not all source tags could
be removed. So my estimate is that we'd actually save less than 1%. Thats
not really worth any effort.

And removing tags actually grows the database because there is now a new
version to keep track of. So its even less worth. :-)

Jochen
-- 
Jochen Topf  jochen at remote.org  http://www.remote.org/jochen/  +49-721-388298