[OSM-legal-talk] Licence Implementation plan - declines or non-responses

Anthony osm at inbox.org
Tue Aug 31 03:22:33 BST 2010


On Sun, Aug 29, 2010 at 10:18 AM, Richard Weait <richard at weait.com> wrote:
> How does one decliner-changeset in the
> middle of a chain of accepter-changestes effect the future data if the
> decliner made one position change, and subsequent editors made further
> position changes?

I'd say usually it shouldn't.  I'd be pretty okay with the following
rules, from a copyright standpoint.

First go through all the nodes:  If a node was positioned in a
particular place by an accepter, keep it, otherwise revert it to the
last accepter-positioned location.  If no accepter positioned it
anywhere in the history, delete the node.

Then go through all the ways:  If a way references two or more nodes,
keep it.  Otherwise, delete it.  Ditto with relations
s/nodes/elements.

Then go through the tags.  Start from the creation of the element.  If
a tag was added by an accepter, keep it.  If a tag created by an
accepter was modified by an accepter, make the modification.

Now, after you've done this there's likely to be some really weird
stuff in the database.  A node might have been reused such that a way
contains nodes on opposite ends of the earth.  Some sort of algorithm
would need to find this type of stuff and delete it and/or tag it for
further review.

Finally, I'd like to mention that some people don't agree with this
algorithm.  They feel an even stricter one should be used.  But as far
as I'm concerned I'd say this is legitimate.  The fact that a POI
merely exists is not and should not be subject to sharealike (*).
It's the positioning of nodes, and especially the positioning and
shape of the ways that should be subject to sharealike.

(*) Ironically, part of the point of the ODbL is to add such
provisions, but without being at least a little bit hypocritical
you're not likely to have much of a database.



More information about the legal-talk mailing list