[Talk-us] Removing tiger:* tags

[Anthony]

On Thu, Jul 29, 2010 at 7:06 PM, Dave Hansen <dave at sr71.net> wrote:
> On Thu, 2010-07-29 at 18:58 -0400, Anthony wrote:
>> Just look in the history for when the way was originally added.
>
> With way combination and splitting, _this_ isn't feasible, either.
> TIGER didn't have any bridges, and so doing this for a road with bridges
> since inserted can be a real pain.

What is the tlid supposed to be for an added bridge?  It doesn't make
any sense.  The tlid is an internal TIGER id.  Way combination and
splitting is exactly the reason I've chosen *not* to keep tlids.
Because tlids don't make any sense once you start changing the ways
from the original data.

> They're defined very precisely in the ruby code that imported them.
> That's publicly available in SVN.  If someone is interested in sticking
> them on the wiki, I'll be happy to point you to it.

Yeah, put in the wiki how the tags represent objective information
about the world we're mapping, and I'll be happy to keep them in OSM.

>> Furthermore, don't store redundant data in the OSM database.  There's
>> absolutely no excuse for having 200 ways which all say name=Cain Rd,
>> name_base=Cain, name_type=Rd.  It's absolutely terrible design.
>>
>> So come up with a better design, define it in the wiki, and I'll keep it.
>
> Heh.  I actually like the concept of separating out the different parts
> of the name.  I think that's a nice design on the part of the TIGER
> folks.

Me too.  But not 200 times for the same name.  And not in a special
"tiger" namespace.

> We can't change the design now.
> It got imported, what, 3 years ago?  I guess we could have a robot crawl
> them like we did the node tags, but what's the gain?

I don't know.  I'm not the one who insists on keeping them.

If you've got data in OSM which is imported and can never be changed,
it shouldn't have been imported in the first place.