On 11/09/19 15:47, Mateusz Konieczny wrote:

11 Sep 2019, 01:54 by pla16...@gmail.com:

    On Tue, 10 Sep 2019 at 23:41, Graeme Fitzpatrick
    <graemefi...@gmail.com <mailto:graemefi...@gmail.com>> wrote:

        Would it need a multipolygon? My impression of the ODA is an
        open patch of ground in / beside a residential area. If that
        is the case, wouldn't it be much simpler to just mark a new
        area in as landuse=o_d_a? (accept it wouldn't be abbreviated)

    Overlapping landuse often works, but only because the carto people
    juggle z-indexes
    to make it work.  They're not overly happy doing that, I believe.

well, reality is that sometimes area
is actually both tree-covered area and
for example university or residential area.

Trees are a land cover, not necessarily a land use.

Universities and residential areas are a land use.

    It also doesn't always
    work well: if ever you've put a pond in a wood without a
    multipolygon you get
    waterlogged trees.

this is intentional to encourage correct
mapping of tree-covered areas.

I wish there was more rendering that showed errors.

      It also makes database queries somewhat more simpler if you're
    asking what is at point A and you get one answer rather than two
    answers, or one of
    two answers chosen at random.

note that in many cases getting two
answers correctly represents reality
Or even 3..
Land cover e.g sand
Land use and e.g. quarry
Land form e.g. dune

      A multipolygon is a little more work for the mapper,
    but not much more.

    Now I expect both the carto and db people to tell me I'm wrong
    about that.  If they do,
    I'll just point out that it's not wrong to use a multipolygon for this

and in even more cases multipolygon
should be used

Many people have problems with multipolygon relations..
Tagging mailing list

Reply via email to