In the Location reports, we have multiple different kinds of "unknown" rows in the country table. E.g. Unknown with EU flag (unknown country in EU), AP flag (question mark) (unknown country in AP region), XX (completely unknown).
So I would assume that we have less "Unknown" rows in the report for "Continent", because in the "Country" report we were able to differentiate the Continents already. But In the Continent report, we have exactly as many "Unknown" rows as all types of "Unknown" from the country report added together.
In the code, I saw why - we are only trying to recognize the continent based on the country we got from the LocationProvider, which fails in a bunch of cases because the country was not provided. So the continent recognition fails even though the provider would have delivered a continent_code in the first place.
I was not able to see why this was implemented like this (though there could be a sane reason, nevertheless). ;-)
Sounds like it could be improved :) 👍 I wonder whether it would put places like "French Guiana" to South America (currently it might assign it to Europe) but these should be edge cases anyway and more correct if it was put into South America