[OSM-dev] Server-side data validation

David Earl david at frankieandshadow.com
Mon Jul 16 13:12:09 BST 2012

On 16/07/2012 12:50, kabum wrote:
> Hi,
> 2012/7/16 Paweł Paprota <ppawel at fastmail.fm <mailto:ppawel at fastmail.fm>>
>     Hi,
>     On Mon, Jul 16, 2012, at 11:07, David Earl wrote:
>      > Yes, it was just an idea. Without buy in for this or something
>     similar,
>      > it won't make progress. It would be easy to implement at the server
>      > side. I had intended to do a prototype, but I ended up with a large
>      > contract to do something else after SOTM 2010 so I wasn't in a
>     position
>      > to. But it won't work as an adjunct - it has to be part of the system
>      > and in particular share the same credentials.
>      >
>      > My thought re the wiki page was that that or equivalent would be
>      > generated automatically from the schema, a bit like tag watch is
>     done,
>      > only a more friendly presentation and, because definitions would be
>      > stored (and in multiple languages), more descriptively.
>      >
>     Tag Central could greatly improve many things but having had some
>     discussions on the topic on IRC I see it's a very delicate topic. On top
>     of that there is the redaction process going on.
>     For now I have started some work in different area just to familiarize
>     myself with the code but this "server-side validation" or Tag Central
>     stuff - that's definitely something I would be interested in. When I am
>     done with some of the UI work, I would love to get back to it, resume
>     the discussion, maybe work on a prototype.
> I would gladly help you with this tool. I'm also interested in the topic
> and maybe we could work together.
> But this week is a busy one and so I maybe read and write more after
> this week.

(If the three object structure is felt to be reasonable) the API is 
generic, with tags and values like the nodes, ways and relations, so the 
actual content doesn't prohibit creating the framework for it.

But beyond that I think the most important thing is to write a more 
detailed spec of the tags and their values. One thing I didn't cover but 
had thought about a bit was how one would say "this thing can be treated 
as this other thing for rendering purposes if you don't understand it 
directly" - e.g. generic sports icon if you don't have one for a 
specific sport, or village in place of barangay.

Had I been able to proceed, my first goal would have been to add enough 
content to be able to reproduce the Map Features page. Then a second 
step would be to add schema editing capability and cross checking to 
JOSM. A third step would be to show how we might have a data driven 
rendering (I know Potlatch 2 has a similar kind of thing, where the 
schema is built in to Potlatch rather than being generic).


More information about the dev mailing list