balrogg at gmail.com
Tue Jan 5 16:17:16 GMT 2010
2010/1/5 Jean-Marc Liotier <jm at liotier.org>:
> As you can see, osm and omp is created roughly parallel in time, but if
> we look back to the root (turistautak.hu) we're "older" than osm.
> And we have different aims as we can see.
> - our primary aim is to create maps for portable gps devices. We have
> routable maps for gps equipments from the beginning (now they are
> searchable as well)
> - we focus on quality, not quantity. Your graph shows that osm have
> millions of track points but do you know about its accuracy and
> actuality? We're trying to create maps as accurate as possible, and we
> do not use every track we get, we use only what we're sure of quality
> about. And a less map area gives a bigger chance to control the
> actuality of map data.
> - we're less wiki-like: not anyone can draw into maps. Map can be edited
> only after a short training course where map editors learn a lot of
> rules how to draw maps. This is for quality, again.
> - osm develops general-purpose technologies (file format, compiler,
> etc). We use existing components where it is possible (mp format,
> mapedit, cgpsmapper, etc..)
> - omp is currently topo and road map as well but hiking-related things
> are more significant while osm is more road-like as we can see.
I see a lot of similarity with a project called UMP which makes a map
of Poland and surroundings, for the Garmin devices, and has all the
properties that you listed:
* It's older than OSM,
* It uses the mp format,
* It has more access control for the Db (their database has a form of
an actual CVS repo with text files - no kidding)
* They use mapedit, cgpsmapper etc.
* They are all about correct routing,
They do however have some wonderfully detailed data with all kinds of
different map features as we do in OSM, the mp format doesn't have
freeform tagging, but they're extending the set numerical codes in the
db practically every other day.
I notice there's some analogy also in their approach to mapping, it's
more practical, fewer dilemmas about the tagging, but also less care
about the "free data" aspect of the project -- while they use
CC-by-SA, they apparently sometimes copy from printed maps and other
sources (though it's hard to tell because there's no source= tag, only
a note here and there), and some of the software they use is Ms
Windows-only -- this is the thing that attracted me in OSM, that the
culture is much like in opensource projects and at the same time 99%
of what we use is also opensource.
So the current situation is they know about OSM, we know about them,
but we don't have real cooperation. There's a consensus that the aims
of the two projects are different. One reason that we didn't try very
hard to join forces is probably that we still have very little data to
offer to them, their map is more complete at the moment.
There are imports being done into OSM, in fact more than half of the
data in Poland currently in OSM comes from them. The situation isn't
clear about what will happen when we relicense, I'm personally still
hoping for some collaboration at some point in the future. There are
also some very low scale imports in the opposite direction, and they
also are using OSM data for some countries other than Poland so that
they can have at least *some* routing there, + a complete mapnik-based
So I don't have any answer to the question of how to join forces with
openmaps.eu. There's one practical suggestion that I'd like to make:
Since openmaps.eu use the mp format, the objects in the source files
probably have no unique IDs of any kind. In OSM every object has an
id. I'd suggest adding an ID tag to the file format and assigning a
unique number to every single object -- the mapedit program should
preserve those tags when the mp files are edited, I think (I've never
used it since it's for windows, but I know the format because I wrote
a converter for OSM). The identifiers are really useful when you want
to merge two databases in some way and I really regret that we haven't
asked UMP to do that. The imports would create some correspondence
between records in the two databases, and any further edits to any of
the two databases could later be automatically applied to the other
one (perhaps only with some simple approval of a human operator).
More information about the talk