[Tagging] Draft Proposal: Default Langauge Format

[Christoph Hormann]

On Wednesday 26 September 2018, Frederik Ramm wrote:
>
> I am thinking of retail outlets like
>
> "Tesco"
>
> "Real"
>
> "Saturn"
>
> "Tk-Max"

Ok, i could of course say now these are not names but brands but 
assuming these are names i see two options for those:

* say the language is defined by where the name was originally 
introduced
* say that these are names without a language (which would in my eyes 
not really be a name any more but an identifying string - could consist 
purely of punctuation or emojiis for example) and integrate these into 
the scheme with a key to be selected.

> But I am willing to extend my definition:
>
> "If the default language of the smallest admin boundary enclosing
> your feature is xx, treat any name tag you encounter as if it was a
> name:xx tag, unless an explicit name:xx tag exists."

By setting the default language of the admin boundaries you would 
retroactively define the meaning of all the pre-existing name tags 
within it without a case-by-case verification.  You would also 
essentially need to forbid changing the default language information of 
the admin units because that would of course break the whole thing.

> That should pretty much catch everything and give everyone the
> flexibility to map all special cases, while not requiring to re-tag
> every single named object all over the planet.

Let me see if i can formulate your suggestion:

check language:default of the feature -> yy

if yy is defined
  name:yy is the name in yy and the local name
  (can be extended to multiple languages)
else
  check language:default of the admin unit -> xx
  if xx is defined 
    if there is name:xx
      name:xx is the name in xx and the local name 
      (and the name tag is meaningless?)
      (can be extended to multiple languages)
    else
      name is the name in xx and the local name
      (with multiple languages semantics are unknown)
  else
    name (if present) is the local name in unknown language(s)
    (legacy fallback)

Apart from the semantic interpretation that name is in xx in case the 
default language is tagged on the admin unit (which comes with the 
problems listed above) this seems exactly the same algorithm i had in 
mind for the proposed method.

The only difference is that because of the problems with the retroactive 
definition of the language of the name tags and the impossibility to 
change the default language of the admin unit in this scenario i would 
in the long term fade out the name tag completely in favour of name:xx.

> I think that generally a more precise recording of languages is
> desirable but if this leads to having to duplicate existing name tags
> into name:xx for every single named place on earth just in order to
> "not defeat the purpose", then I'd like to pass on this idea and wait
> until a less intrusive one comes along.

As said the aim is not to duplicate the name data but to replace the 
generic name tag that has no defined language with the individual 
language name tags.

-- 
Christoph Hormann
http://www.imagico.de/