search

UW-Macrostrat / macrostrat-api

The API for SCIENCE
3 stars 1 forks source link

NULL entries for GPlates rotations #149

Open jonhusson opened 8 years ago

jonhusson commented 8 years ago

is it possible to have the field empty rather than '0' for _plat and _plng where no rotation exists? It is a little confusing but not unmanagemeable

https://macrostrat.org/api/units?response=long&unit_id=240,379,380,620,1045,1046,1048,1527,1648,1686,1687,1846,2337,2341,5117,5399,6650,8745,8746,8778,9103,10359,10687,10747,11687,12824,12849,12850,12853,12948,12986,12991,13573,15510,15816,17338,17359,17360,17364,17365,17429,17443,17445,17994,18052,18176,18178,22788,34160,34496,34532,35827,35864,37440

jczaplew commented 8 years ago

Definitely. 0 is a terrible NULL when dealing with coordinates, as it is a valid coordinate.

Would we consider this change breaking?

jonhusson commented 8 years ago

I think it probably would be breaking, but I still vote for making the change to something thats more obviously not data (or not model output, rather). Especially if people only focus on paloelatitude (which I suspect will be many users), a value of 0 as NULL has a high potential to be missed

jczaplew commented 8 years ago

Agreed. This falls into the data quality category. @cambro any objections?

cambro commented 8 years ago

No objections. However, this should not be a breaking change. NULL is appropriate.

jczaplew commented 8 years ago

NULL isn't an option, as it doesn't produce valid JSON. I suggest -9999

cambro commented 8 years ago

Ugh, that's annoying. I don't like having a number there at all in this case. Is it totally non-kosher to have the string "NULL" or '' in cases where there is legitimately no data?