[OSM-talk] Mapping with no GPS data

David Earl david at frankieandshadow.com
Tue Oct 17 11:07:47 BST 2006



> -----Original Message-----
> From: talk-bounces at openstreetmap.org
> [mailto:talk-bounces at openstreetmap.org]On Behalf Of Thomas Walraet
> Sent: 17 October 2006 10:45
> To: talk at openstreetmap.org
> Subject: Re: [OSM-talk] Mapping with no GPS data
>
>
> Nick Hill a écrit :
> >
> > Perhaps we should work on the page:
> > http://wiki.openstreetmap.org/index.php/Starter_Tutorial
>
> This page told to beginner to just make segments. It may be a good idea
> for simplicity, but the fact that osmarender don't display segment by
> default could be a problem.
>
> I think osmarender should display segments by default (ie in the
> preference file example). To encourage way-creation, segment that don't
> belong to a way could be drawn with red dotted line. This will give
> visual feedback about an area progress, and stop people wondering why
> their maps are empty.

I don't think this is a good idea. It means that people don't finish the
job. Segments are OK as an interim, but they're not suitable for attaching
steet names and the like to.

There have been hints that there's been a discussion about abolishing
segments, before I joined. This seems to me to be the way to go, though
you'd have to be careful not to lose data. But the segment tool in JOSM and
the online applet (does this have a name?) would create ways linking nodes
directly.

The downside of this is that abutters and other partial-way data can't be
represented. I think this could be done by allowing properties to be
attached either to a whole way or the parts of a way. In effect implicit
segments. But they need only be stored for those parts which need them, and
can be represented the "other way up" from segments now, e.g. in mysql as a
table with wayid, partid, startnode, endnode and in the xml as a subordinate
part of the way, as I posted recently.

Alternatively and more simply, upwardly compatible, and less work, retain
the segment, but suppress it in the user viewpoint. So JOSM would never show
the segments that it implicitly creates when a way is created, but allow
selection of parts of ways (with a different tool ideally, rather than the
awkward ALT-CLICK), which would in effect select the implicit segment.  Not
sure what would happen to segments which aren't now in ways - maybe show
them in a separate layer, like tracks are, so that they can still be used
for tracing proper ways, but they aren't regarded as part of the final map
data. For backward compatibility, when ways are traced, properties already
assigned to the segments could be transferred to the way by default, but for
new data this would get done properly.

There's a halfway house between the two which might improve efficiency:
convert Ways to sequences of node pairs eliminating segments from way
definitions, but retain the segment data for this separate layer. The ways
could then reference those segments which it needs to deal with local data
like abutters, bridges and tunnels etc.

Incidentally, I think it should not be permitted to upload or enter a Way
which doesn't at least have a type of some kind set for it (e.g.
highway=...).

David





More information about the talk mailing list