[osmosis-dev] Duplicate keys in replication diffs

[Paul Norman]

On 2020-09-16 11:04 p.m., Stephan Knauss wrote:
> Hello Paul,
>
> Thanks for tracking down and reporting this bug.
>
> On 17.09.2020 01:22, Paul Norman via osmosis-dev wrote:
>> In changeset 90486873 a node was introduced with the tags
>>
>>     <tag k="shelter_type" v="weather_shelter"/>
>>     <tag 
>> k="shelter_type^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?^?" 
>> v="lean_to"/>
>
> You did not mention: Have you already filed a bugreport against 
> potlatch2? I think it was not intended to push this key.

I mentioned it to RichardF, but have been following up on the 
replication side first.

> Have you checked whether we have more unintended tagging like this in 
> the database?

We've encountered about 50 other occurrences of duplicate keys which are 
likely caused by this bug. I have no estimate on how many keys in the 
database have control characters without conflicting with another tag.

> Taginfo has a report about potentially problematic characters, but it 
> does not seem to consider control characters.
> https://taginfo.openstreetmap.org/reports/characters_in_keys#problem

The OSMF taginfo instance uses the replication diffs, so is impacted by 
the osmosis bug like any other diff consumer.

For next steps, I'll write this up for the osmosis issue tracker.