Open Radar

 
Summary:

Geocoding a Lat/Lon to address via the CoreLocation API yields one or more CLPlacemark objects with "areasOfInterest" array.

This value is poorly documented but appears to yield an array of strings, which can vary from the examples of a store name or retail park / municipal area all the way up to an entire country.

This makes the data next to useless as there is no classifier on the scope of each area of interest in the list.

For example if I want to show the nearest place of interest by name, e.g. "Apple Store", I cannot rely on this info because, for example, at my address in the UK (outside Stroud, Gloucestershire, England) where there is *no* area of interest in Maps nearby, areasOfInterest contains "Great Britain".

I can't show the user "You are at: Great Britain, Brimscombe". But I *do* want to tell them "You are at: Apple Store, Bristol" if there is a *real* area of interest.

Steps to Reproduce:

1. Use CoreLocation to look up a lat/lon to get CLPlacemark instances
2. Access the areasOfInterest property of the placemark and notice the disparity of data
3. Rinse and repeat in locations where there is no significant area of interest data / use my example location

Expected Results:

areasOfInterest should be empty if it would only contain a large-scale area such as an entire country.

Actual Results:

areasOfInterest contains e.g. "Great Britain" which is somewhat overshooting the "areas of interest" concept for a given lat/lon. One might argue that the nearest city is a valid area of interest, but really this appears to be meant to represent PLACES that are nearby.

Regression:

N/A

Notes:

Unless I'm using the API wrong, it looks like areasOfInterest is really completely unreliable.