[OSM-dev] Fail to get OsmChange for some changesets

Andy Allan gravitystorm at gmail.com
Tue Oct 5 11:49:45 BST 2010


On Tue, Oct 5, 2010 at 12:49 AM, Manchun Yao <manchuny at microsoft.com> wrote:
> I am learning how the change set works in OSM.  I assume that if I can apply
> the change sets of a particular area one by one then I can obtain the
> current osm map of that area. Is this a valid assumption?

Not really. Changesets don't affect the database in one go, so you can
have two changesets affecting different objects in non-sequential
orders. For example

Changeset 1: Node 10v1 Node 11v2
Changeset 2: Node 10v2 Node 11v1

You can see there's no order of applying these changesets that will
leave your database with both nodes on v2. This is very rare in
practice, but it might trip you up at some point.

> I picked the Boulder, Colorado area since the number of changes there,
> roughly 350,  is about right.   During the tests, I noticed that I could not
> obtain the OsmChange of the following changes: 145015, 4318130, 4318132,
> 4318136.

You're going to have to explain further what you're doing, but it
doesn't sound like you are editing the data, so making repeated
requests to the editing API isn't the best way. Are you trying to
build a history of all the changes for the area? Or are you trying to
end up with the current state of the map?

Cheers,
Andy



More information about the dev mailing list