I neglected to look at the namespace before adding the :: calls for all internal dependencies. Thought it might be worth considering which approach was preferred. I know there is a little latency for ::, but it can also clean up the namespace a little bit.
Add tbl_df to the class attribute of a tbl_json object. This makes tbl_json a true extension of a tbl_df (and thereby cleans up a lot of the internal function calls). Also, the "as_data_frame" or "tbl_df" functions were often used internally to drop the JSON attribute, drop the tbl_json class, and convert to a tbl_df, so I made this explicit
Fix DROP behavior to be consistent with a tbl_df. Namely, disregard DROP=TRUE and print a warning message that it was ignored. Note that this is backwards-compatible, as DROP=TRUE caused an error previously, because DROP=TRUE ceases to be a data.frame, and as such ceases to be a tbl_json. i.e. it fails in the assertthat for as.tbl_json
This PR accomplishes a few things:
::
calls for all internal dependencies. Thought it might be worth considering which approach was preferred. I know there is a little latency for::
, but it can also clean up the namespace a little bit.DROP=TRUE
and print a warning message that it was ignored. Note that this is backwards-compatible, asDROP=TRUE
caused an error previously, becauseDROP=TRUE
ceases to be a data.frame, and as such ceases to be a tbl_json. i.e. it fails in the assertthat for as.tbl_json