[Openstreetmap-dev] OSM's Schema - moving it forwards.
Tom Carden
tom at tom-carden.co.uk
Mon Nov 28 23:27:31 GMT 2005
Andy Robinson wrote:
> I spend a lot of my working day (when not distracted by OSM ;)) writing
> technical specifications. When doing so, the clever approach is not to
> prescribe what shall or shall not be permissible but what the performance
> should be. What I'll throw out to you all shortly for comment will be a
> performance based schema not a prescriptive one
OSM has always been driven by the Simplest Thing That Could Possibly
Work philosophy, and the requirement that all changes be logged in that
beloved, "wikified" way. Though I just about agree with Nick that a few
things are so ubiquitous as to be worth putting in the schema, I like
the idea that there is a single field for metadata as it makes tracking
changes super simple. Adding more columns complicates things, adding
dedicated fields when a general one will do just adds complexity. Would
it actually let you do more stuff, make more expressive maps, undermine
more monopolies, draw prettier maps, or are you just being
perfectionists? The tags are a little clunky, but they work, and
parsing them is trivial, if not elegant.
My vote would be to move along on the schema talk, and worry about some
of the hard problems like representing changes in the client, rolling
back changes in the client, exporting to useful formats etc etc etc.
OSM's job as far as Imi described it has to be C - it's both an API and
a web client. And the web client is the number one consumer of the API,
so the pay-off in writing it in this way is clear. If the project
doesn't lead from the front with sensible defaults (e.g. name tag for
street name) then nobody else will.
The future of the client as Ben was describing it (AJAXy goodness) is
inevitable I think - IE could do it now with VML, and the imminent
Firefox 1.5 and Safari can do it with either Canvas element or SVG. The
latter would be totally Free-as-in-Freedom, and allow us to ditch
reliance on plug-ins.
There's about 10p in the pot so far, I think that's enough :)
Best,
Tom.
More information about the dev
mailing list