[OSM-talk] Validator tags

Matthias Julius lists at julius-net.net
Fri Nov 21 18:07:41 GMT 2008


Pieren <pieren3 at gmail.com> writes:

> Taging for validation is already a controversial subject.
> But some of your tags are even more controversial:

There are two levels of "tagging for the validator" (or any other
application):

- adding validator specific tags to give the validator a hint what to
  do

- abusing existing tags to get the validator to do the "right" thing

With the first I agree.  It is just not possible to define validation
rules that always produce the right results.  How is the validator to
know that this street really has no name?

The second one is evil.  An example could be to tag "name= " to get
the validator to shut up without messing up rendered maps.

>
>> validate:empty-tag-key
>> validate:empty-tag-value
>
> A tag with an empty key or an empty value is not possible/ not
> allowed. Simple delete or fix the tag if you find such. Don't say to
> maplint "well it's wrong, but don't care about it".
>
>> validate:untagged-way
>
> Again, it's nice if your validator shows untagged ways. But then
> contributors must fix them either by setting the correct tag or the
> highway=road or deleting them if they are too old. The aim of the
> validator is to clean-up the database, not to clean-up the map showing
> errors.
>
>> validate:deprecated-tags
>> validate:motorway-without-ref
>
> Same thing : if maplint complains, fix the error by fixing the mistake
> in the osm db, not by adding a tag for maplint. If a tag is
> deprecated, replace it by the new tag !
>
>> validate:poi-without-name
>> validate:residential-without-name
>> validate:place-of-worship-without-religion
>
> Well, this is an extension of the noname issue, now. I'm waiting the
> next one complaining about the missing
> "validate:highway-without-maxspeed",
> "validate:highway-without-oneway",
> "validate:highway-without-maxheight",
> "validate:highway-without-lanes", etc..

In general, there should only be ignore rules for specific tags that
have a "legitimate" use case.  If there are highways that really don't
have a name maplint should not complain about it.  But, if it is not
possible that there is a place_of_worship without a religion it should
not be possible to suppress the maplint warning if it has none.  If
the relegion is simply not known this is still an error in the data.

Matthias




More information about the talk mailing list