Just a some additional comments on my behalf:
I noticed that very complex multipolygons are increasingly used throughout Switzerland.
In conceptual terms I fully endorse their use since it seems currently to be the mapping method with the least redundancy. However, understanding the objects and even more so editing these highly relation-based objects seems considerable more complex to me than using somewhat overlapping ways, and I am afraid that this mapping method puts a considerable, additional barriers for casual and new mappers. To me a large and lively mapper community is key for osm's success and thus mapping should be kept as simple as possible. Thus, the whole question boils down to how we want to balance two somewhat diverging goals: redundancy-free data vs. simple mapping practices leading to a wider mapper community.
When discussing the question of multipolygon vs. overlapping ways we already assume that the objects share some nodes. On the talk-de (thanks to ze german preference for Grundsatzfragen) there were already a couple lengthy threads (the latest see [1]) on whether this should be done at all. The main main arguments are: (i) the ease of understanding and editing the objects and (ii) the problem that some for some objects we are mapping not the outline (highways) while for others we do (landuse). If sharing nodes is sought, then there are two methods: using multipolygons or multiple ways partly overlapping.
Currently most of our data is in ways that are directly tagged. It seems fairly easy as well as handy to transform (some) of them into complex multipolygons. However when I want to add or edit something that is already mapped in the "complex-multipolygon-method" this seems quite a challenge just to understand what's already comprehended into the data. When then making edits, I am somewhat never certain not to have broken any relations. One of my main concerns is, that the ways included into the multipolygons get shorter and shorter with every additional multipolygon (shorter segments) resulting in a even less human-understandable data structure. With the mapper Sue Average or Joe Beginner in mind I am somewhat skeptical he or she (i) understands the data, (ii) dares to make any edits and (iii) edits the data so no relations are broken.
Possibly this will all be resolved with further development of the editors. But for the time being a widespread preference of the complex-mulipolygons-method seems problematic regarding our mapping community.
Best regards
Marc
[1] http://lists.openstreetmap.org/pipermail/talk-de/2011-August/088266.html
On 16.08.2011, at 23:10, Stéphane Brunner wrote:
Hello,
With FischersFritz we have a discutions about where using Multiyolygone.
Personally I relay like using multipolygon for many reason, essentially because I many times tine find some area where the polygone have some commons points, an some unwanted space or some overlapping parts, when we use multypolygone is clearer, and we make less mistakes.
An other reason ti when we have some overlapping ways (sometimes more than 6) ti becomes relay difficult to select the right way.
But the main disadvantage of using multipolygon is that's a bit more difficult to understand. than we can have some thing that that (start of the discussion): http://forum.openstreetmap.org/viewtopic.php?pid=184210#p184210
An other point is that if a multypolygon isn't complete his display can be not so good in Josm.
Than for me multypolygon should be use if we have two area with a common long border to say something with more than 10 points in a row (rules that from now I didn't respect ;-) ).
FischersFritz have you something to add ?
What's your opinion ?
CU Stéphane
-- Envoyé depuis mon lapin -- Catalogue de cartes OpenStreetMap - http://map.stephane-brunner.ch Un peu d'espace qui vous suis partout - https://www.getdropbox.com/referrals/NTk2OTU2Mjk -- _______________________________________________ talk-ch mailing list talk-ch@openstreetmap.ch http://lists.openstreetmap.ch/mailman/listinfo/talk-ch