You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This library should add the Swift port of Turf as a dependency. GeocodeResult and much of the Placemark struct’s Codable conformance should be replaced by FeatureCollection.
Rationale
This library has a significant amount of response parsing code. Even parsing something as simple as a set of routable points is a hassle:
The existing code is functional, but it requires manual updates any time the Geocoding API changes. Supporting a companion API such as the Japan Search API requires extra care just to get something functional.
As it happens, both APIs’ response formats are valid GeoJSON by design. The Swift port of Turf contains full-fledged, well-tested GeoJSON parsing functionality and model types. MapboxDirections successfully migrated to Turf for its geometry-related needs in mapbox/mapbox-directions-swift#382.
This library would continue to be a lightweight, maintainable wrapper atop the search APIs for use cases that don’t require the full heft of the search SDK or that require specialty endpoints like batch geocoding. Conversions to other standard iOS libraries like Contacts would remain, but the smaller codebase would make it easier to support additional Swift platforms like Linux and Windows with negligible overhead.
Implementation notes
Much of the metadata for each result is stored in GeoJSON foreign members, which are now supported by Turf: mapbox/turf-swift#175.
Turf doesn’t bridge backwards to Objective-C. In order for this library to retain Objective-C support, it would need to wrap any Turf types in Objective-C-compatible classes. That might not be much of a problem in the short term, since the response parsing is mostly an implementation detail. However, to achieve the long-term maintainability gains described above, we may want to limit Objective-C support to the current functionality or even drop it in future versions.
This library should add the Swift port of Turf as a dependency. GeocodeResult and much of the Placemark struct’s Codable conformance should be replaced by FeatureCollection.
Rationale
This library has a significant amount of response parsing code. Even parsing something as simple as a set of routable points is a hassle:
MapboxGeocoder.swift/Sources/MapboxGeocoder/MBPlacemark.swift
Lines 453 to 460 in 9dcd38e
The existing code is functional, but it requires manual updates any time the Geocoding API changes. Supporting a companion API such as the Japan Search API requires extra care just to get something functional.
As it happens, both APIs’ response formats are valid GeoJSON by design. The Swift port of Turf contains full-fledged, well-tested GeoJSON parsing functionality and model types. MapboxDirections successfully migrated to Turf for its geometry-related needs in mapbox/mapbox-directions-swift#382.
This library would continue to be a lightweight, maintainable wrapper atop the search APIs for use cases that don’t require the full heft of the search SDK or that require specialty endpoints like batch geocoding. Conversions to other standard iOS libraries like Contacts would remain, but the smaller codebase would make it easier to support additional Swift platforms like Linux and Windows with negligible overhead.
Implementation notes
Much of the metadata for each result is stored in GeoJSON foreign members, which are now supported by Turf: mapbox/turf-swift#175.
Turf doesn’t bridge backwards to Objective-C. In order for this library to retain Objective-C support, it would need to wrap any Turf types in Objective-C-compatible classes. That might not be much of a problem in the short term, since the response parsing is mostly an implementation detail. However, to achieve the long-term maintainability gains described above, we may want to limit Objective-C support to the current functionality or even drop it in future versions.
/ref mapbox/MapboxStatic.swift#112
/cc @OttyLab @tsuz
The text was updated successfully, but these errors were encountered: