piccolbo / dplyr.spark.hive

spark and hive backends for dplyr
8 stars 1 forks source link

head and collect don't use same conversion mechanism #6

Closed piccolbo closed 8 years ago

piccolbo commented 8 years ago

As discovered while collecting a table create from a json file with some degree of nesting, collect fails whereas head doesn't (but uses character for a map, which isn't good either). Best would be to unify the two code paths and support as many hive types as possible. Complex data frame columns should not be a priority though as their use is limited.