[Talk-lv] Riverbank tagging

[Tomas Straupis]

> I skimmed though the doc. And it seems that the argument is about using waterway=riverbank vs water=river for water covered areas of a river.
> Assuming both are rendered the same, is there any practical difference between the two?

  Using waterway=riverbank forces mappers to distinguish
lakes/reservoirs from riverbanks which is important for:
  * symbology  - waterbodies like lakes/reservoirs and riverbanks are
two different classes of objects and are symbolised differently,
starting with labels where waterbanks do not need any labels but also
with object usage on different scales where waterbank would always
have a corresponding river while lakes/reservoirs. And from practical
perspective, I've NEVER ever was in a situation where I would want to
symbolise riverbanks the same way as lakes/reservoirs.
  * analysis - once again riverbanks are totally different beast from
lakes/reservoirs and are never lumped up together for analysis (still
water vs flowing water).

> On the surface it seems a very much clinging to a name because that is how we started out 10
> years ago and the community process was not great about the change.

  BTW that is also an important aspect. For each change we have:
  * B - benefit of a change
  * C - cost of a change
  Where cost of a change is a sum of:
  * cost of a technical change (updating the database - the cheapest one)
  * cost of related things (codes, maps, qa procedures etc.)
  * cost of updating documentation (documentation, training materials,
books etc.)
  * cost of mindchange (things stick into our minds:-)
  For any change to be beneficial, B must be higher than C.
  The only benefit of this new water schema is for coders - writing
one where condition "natural = 'water'". And paradoxally  - this is
usually an indication that a coder is doing an incorrect query as as
stated above these objects should never be lumped up together.

> IMO water=river fits better with the overall tagging pattern.
> waterway=* is about the same as highway=* and used for adding routable data and centerlines.
> The similarly used but not officially approved area:highway=* for highways. I would personally prefer area:waterway=* just for consistency sake.

  Well changing waterway=riverbank to area:waterway=* would be fine
with me as that would keep the explicit distinction between the
classes in question.
  Lumping everything under natural=water is the main problem.

> And maybe try to focus on proposing a better tagging scheme going forward.

  Well in order to fix something there must be a problem first, right? ;-)

-- 
Tomas