[OSM-dev] osm2pgsql for 64-bit IDs

[Jochen Topf]

On Sun, Jun 05, 2011 at 08:28:30PM -0400, Anthony wrote:
> On Sun, Jun 5, 2011 at 7:38 PM, Scott Crosby <scott at sacrosby.com> wrote:
> > On Sun, Jun 5, 2011 at 8:41 AM, Anthony <osm at inbox.org> wrote:
> >> On Sun, Jun 5, 2011 at 8:39 AM, Frederik Ramm <frederik at remote.org> wrote:
> >>> Linestrings with internal geometries are very unlikely to happen, for a
> >>> number of reasons, one of them being topology.
> >>
> >> That's too bad.  More than 90% of node ids are completely useless as
> >> they refer to only one way.
> >
> > Out of curiosity, I did a test on this, because I was planning on
> > exploiting this 'conventional wisdom'. I was hoping that 99% of nodes
> > occur in one way, but on a 2010-08 planet, 85% of nodes occur in one
> > way, 12% in 2, and 1% in more than two..
> 
> Yeah, last time I checked it was just slightly over 90%.  But even if
> it's only 80% of nodes that are completely useless, I think it's worth
> doing something about.  Or at least considering doing something about,
> and not writing it off as "it has to be that way but I'm not going to
> tell you why".

So let me see whether I got this right:

You have some vague idea how you are going to save us a lot of space by somehow
storing coordinates in ways. You think its "worth doing something about", but
you haven't done anything. You haven't explained how exactly this is supposed
to work in detail. You haven't explained how you are planning to change every
single piece of software that handles OSM data. You haven't explained how
corner cases are handled such as when all tags are removed from a node in a way
(does this node then loose its ID and is integrated into the way?). You haven't
explained how we would get there from where we are.

You haven't done anything on this idea *you* came up with. But you expect all
the other people who don't believe in this idea to do the work? How are others
supposed to explain the problems with your plan if you haven't even got a plan,
just a vague idea?

Please, if you think this is worth working on, then do so. Work on it. Study
the problems, change some software, come up with a plan, try to implement it,
find flaws, rework the plan. Do something. 

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