[OSM-dev] New OSM Test Data Repository

[Jochen Topf]

On Fr, Feb 21, 2014 at 08:05:34 +0100, Christoph Hormann wrote:
> On Friday 21 February 2014, Jochen Topf wrote:
> > I have started to work on a common OSM test data repository that can
> > be used to test all sorts of software that works with OSM data. See
> > my blog post at
> > http://blog.jochentopf.com/2014-02-21-osm-test-data-repository.html
> > for details.
> 
> Nice.
> 
> Of course I could not resist creating another test:
> 
> http://www.imagico.de/files/data.osm
> 
> where the result would be 'valid but invalid in plate carree or web 
> mercator'.
> 
> 'invalid but valid in plate carree and web mercator' is also possible of 
> course, just like all other combinations. ;-)
> 
> In case you wonder about the coordinates - that's (500000 9000000) and 
> (550000 9050000) in UTM zone 47.  And of course this does not work near 
> the equator.

I am not sure yet what exactly the "valid" and "invalid" mean and whether
we need different outcomes. For the moment "valid" means more something
like: "The data looks good, could exist in a real database, and everybody
would agree on how to interpret it." While "invalid" is everything else.

Say, for instance, that you have a way that contains the same node twice
directly next to each other. Do you remove one occurence in the linestring
geometry or not? You don't see the difference usually, so in almost all
cases it does not matter. But there could be a difference. And it doesn't
happen often in the data. But strictly speaking people shouldn't add data
like that. So is that already "invalid" data?

It also depends on what you are testing. If you are testing the OSM XML parser,
those cases are valid data that the OSM parser must understand. If you are
testing the renderer, it might be different. I expect different programs
to use the same data.osm files from the repository, but decide for themselves
which test case is valid and which isn't in their context.

I haven't figured out all the details yet, for the moment I am trying to
get the ball rolling, get started writing test data and tests programs that
use them. We'll figure out the details along the way.

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