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
For programming, it would be nice to allow for vectorized paths. At present, it is possible to workaround with do.call, but this does not work nicely in a pipeline.
I think a useful (and tidy) way of implementing would be the tidyeval framework that was recently added to the rlang package.
Bonus : If it is possible to somehow vectorize (or automate) type-determination (like the readr package does), I believe that would make for a nice solution to the typing required for spread_values!
The text was updated successfully, but these errors were encountered:
Determine names, handle missing or duplicates... Maybe handle recursion first to get all possibilities for consistent naming, rather than deduplicating while recursing
Generate json_spec (requires json_spec class). Some methods for this: json_skip,json_factor,json_dbl,json_chr,json_int,json_lgl,json_date,json_datetime,json_only... naming convention up for grabs using the purrr or the readr conventions
Execute json_spec to spread out the JSON as designated.
The hope is that this could be combined together into a single spread_json function which has nice guess behavior like spread_all, but is also easily standardized like spread_values, but with less typing.
For programming, it would be nice to allow for vectorized paths. At present, it is possible to workaround with
do.call
, but this does not work nicely in a pipeline.I think a useful (and tidy) way of implementing would be the
tidyeval
framework that was recently added to therlang
package.readr
package does), I believe that would make for a nice solution to the typing required forspread_values
!The text was updated successfully, but these errors were encountered: