[OSM-talk] Proposed new status for tags in the wiki: "import" for undiscussed tags that were only used by an import
Roland Olbricht
roland.olbricht at gmx.de
Thu Mar 19 05:37:10 UTC 2020
> I would like to add a new status to
> https://wiki.openstreetmap.org/wiki/Approval_status and to Tag and Key
> pages in the wiki
>
> The value "import" would be used for tags that were used in a large
> import, but are no longer commonly being added.
The misunderstanding about "object:street" has already shown that things
are probably not that easy. The tag is in widespread use, but just not
on the entire planet.
In other words, that status "import" would conflate at least three very
different things:
"local": tags that are in use by a local community. There are many more
subtle things that are important locally but do not make sense on a
global scale. The "object:street" is an example of this, or
"addr:conscriptionnumber" mostly in Austria. "crossing_ref" is an
example for an UK specific tag, GNIS are local to the US and so on.
"external": There are quite a lot of references to an external databases
(e.g. "ref:bag", "osak:identifier", "wikidata", IFOPT and NaPTAN). They
may or may not be global. In any case, they are meaningless unless you
have access to the source database as well, as opposed to "local" which
are verifiable on the ground.
"stale": Tags that came with an import, are not, and can not be used by
general mappers, and are not expected to be updated. "tiger:cfcc" is
currently the most numerous. The low number of values of "tiger:cfcc"
makes it unlikely that it is carrying any meaning.
Another final question is whether it makes sense to refine the system at
all. Much of the information of the tagging classes is available via
taginfo, and some more can be automatically computed from the database,
although it is it done today. Having this is as explicit information
instead is the either redundant (if in line with actual database
content) or misleading (if in contradiction to the actual database content).
Best regards,
Roland
More information about the talk
mailing list