[OSM-talk] Re: [OSM-dev] MassGIS dataset

[SteveC]

I'm back home, and I have internet, woo!

* @ 03/06/06 07:32:58 PM crschmidt at crschmidt.net wrote:
> > Oh yes you can, for a small area you can use the API.  It's a 
> > plain HTTP get request that returns the same kind of XML you get 
> > in the planet.osm dump.  This is enough for a proof of concept.
> 
> What do you consider a usable proof of concept? A single block? A square
> mile? A single degree by a single degree? I don't have any interest in
> scales less than statewide -- and I can't even get a half-degree by a
> half-degree API query out of San Francisco in less than 3 hours. (As of
> yesterday morning.)

What you and others are doing (attempting to get large amounts of data
out) should be something that just works. Really. You should be able to
get up to date data out of it, or have daily dumps. But we're not there
yet for social, financial and technical reasons which are interlinked
as Andy explained or you pointed out.

So, I've brought down the database timeout from 8 hours to 4 minutes as
these massive area queries spoil the party for everyone. This isn't an
invitation to make millions of < 4 min queries.

I'll revise the timeout value with feedback.

> Since there's been a request *not* to grab large chunks of data via the
> API, what exactly do you expect would be a decent demo? I'm glad to give
> you a demo to do whatevery ou want with, but for my needs, OSM can't
> serve them at the moment.

As far as I know your needs (openlayers using OSM data for where 2.0?
I've had to skim email) don't intersect well with most other OSM users.

You're in the US and have access to data anyway, most OSM users don't.
You want lots of data out, most want to put little bits in, etc.

NickW's rendering work does intersect because it improves the rendering
of the existing view/edit/upload cycle. Integrating openlayers instead
of using the hacked anselm tile.js would be great too, or improving the
node query (which I'm sold on, it just needs the table locking /
transaction work, I'm not convinced that querys won't clash without
that) is cool.

The specific pledge you raised is poor from the social angle, it's not
the way things should go. From the technical angle, it'd be great to
have another db machine but from the financial angle it's chicken and
egg to fulfill your request.

What's great is despite all these denials to things you want to do, you
still want to work on stuff to get there. So, if there's any way to
help, then here I am. #geo really isn't the place to ask deep questions
unless mikel or I are on.

have fun,

SteveC steve at asklater.com http://www.asklater.com/steve/