[OSM-talk] [OSM-dev] Latest planet.osm contains incorrect data

[Petter Reinholdtsen]

[Steve Coast]
>> I guess the intended behavior is to store UTF-8 in the database and
>> export UTF-8 in the dump.  For that to work properly all the database
>> entries with ISO-8859-1 will need to be converted.  It should be
> 
> Actually no, because the API deals things out correctly.

I'm a bit confused by this comment.  Do the database store the strings
as UTF-8 or at least using some encoding and reportoare with the full
Unicode/ISO-10646 reportoar?

If it does, why does it matter what mysql can convert?

> MySQL can and does on-the-fly collation and spits out UTF8 correctly
> on the API. The puzzle is why (using exaclty the same code on a
> different machine) it doesn't in the planet dump. I have yet to
> check the mysql client versions are the same etc, I will now. (This
> we before the explicit set utf8 call patch to the planet dumper).

And if it does not, how can we make sure all characters can be stored
as street names?

Friendly,
-- 
Petter Reinholdtsen