Issue 103 - geojson_sf() with mixed logicals

[ateucher]

This is an attempt to fix #103.

In get_property_types(), property_type ("True" or "False") essentially just reflects which boolean value was encountered first, but subsequent features can have a different type, so that type for a feature can be different from property_type for that property. Before, if a property had different types across features, it was immediately converted to "String". I added logic to recognize that "True" and "False" are the same logical type.

In fill_property_vectors(), it now allows both true and false values to be stored in the logical vector, regardless of whether the detected type was labeled as "True" or "False".

I added some new tests, which now pass (failed before), but definitely warrants checking my logic!

[dcooley]

Thanks @ateucher, looks good to me! Could you add (if you want to) a contributor role in the Description for yourself?

[ateucher]

Great, thanks! Will do!

[ateucher]

@dcooley do you want me to update the GitHub Actions workflows? They're not running as they use old unsupported workflows. I think r-lib/check-standard should be sufficient...

[ateucher]

Hi @dcooley just a little nudge on my question...

[dcooley]

sorry, am away on holiday; yes if you could get the actions working again that would be great

[ateucher]

Ah, my apologies! Will do :)

[ateucher]

I made some updates to the DESCRIPTION file, as installing the dependencies was throwing a dependency conflict error (I think since jsonify is listed in both LinkingTo: and Suggests: and only once with a minimum version... and that minimum version requires the GitHub version as specified in Remotes:

I guess before this could be released to CRAN, jsonify, rapidjson, and sfheaders would all need to be on CRAN...

[dcooley]

excellent, thanks for those changes. Yes I'll update the other packages hopefully in the next couple of days. Then all should be good to go for CRAN