[OSM-legal-talk] sharealike trigger
penorman at mac.com
Tue Jul 23 18:43:57 UTC 2013
> From: John Bazik [mailto:me at johnbazik.com]
> Subject: Re: [OSM-legal-talk] sharealike trigger
> On Mon, Jul 22, 2013 at 10:46:30PM -0700, Paul Norman wrote:
> > What do you mean by fields?
> I mean columns in RDBMS tables.
I don't believe you can make any general comment about columns
in a RDBMS table. For example, you could use a RDBMS as a k-v
store and store both OSM data and a completely unrelated set
of photos in it. On the other hand, you could use a schema which
is in 3NF and then the OSM database is over many different tables.
As a practical matter, most data consumers probably want an
un-normalized table because they are only reading.
I had thought about saying something about tables in un-normalized
forms, but I'm not sure if it's general enough.
I still believe it's best to avoid talking about a technology-specific
way of storing databases, particularly when the OSM database comes as
an XML file, which is closer to a set of flat text files than an RDBMS
in many ways.
I also found when I shifted my thinking to avoid using any terms from a
particular way of storing databases the database directive became clearer.
> > One description for the OSM map database (planet.osm) is a database of
> > georeferenced shapes (including points) with associated data (what the
> > shape
> > represents) and meta-data (time edited, user edited by, etc). I am
> > very
> And OSM excludes *some* of that meta-data from what it considers to be
> derived - "fields" like "user edited by."
I'm not sure what you mean. The meta-data included in planet.osm is part of
the OSM map database, although most consumers drop the user-related metadata
because it unnecessary for most applications.
> In RDBMS terms, might one expect to use a SELECT to produce a derivative
> "database," excluding non-derivative data?
Well, in RDBMS terms SELECT produces a rowset which is closer to a table,
not a "database".
More information about the legal-talk