Currently residential landuse for towns/cities etc. seems to be drawn over the top polygon in it’s own right. For cities where this hasn’t been done, is it reasonable to create a multipolygon and use existing ways (where appropriate) to mark the edge of the city?
I see the advantage of using a multipolygon is that it reduces the number of extra/duplicated/parallel ways that need to be drawn etc. The disadvangtes are adding new bits to the relation might be a bit of a pain and there’s the potential to accidentally break things when remodelling areas.
Using a multipolygon appears to render ok, I’ve tried it for a small area where I map - http://www.openstreetmap.org/?lat=51.4452&lon=-2.4609&zoom=13&layers=B000FTF . Using this approach does mean you have to split things like roundabouts that fall on the perimiter.
Before I “go crazy” and mark the boundaries of Bristol, Bath etc. I was wondering if there was a “correct” method.