It may be useful to allow ingesting legate-dataframe tables even if this is mainly for convenience.
Right now, we do provide some helpers to convert to cupynumeric via table.to_array() already, so this may be mainly a short utility to call these automatically.
There may be a use/need to automatically cast columns into a common dtype, so that this may be more of an issue for legate-dataframe, where we would want to support to_array(dtype=, null_value=).
It may be useful to allow ingesting legate-dataframe tables even if this is mainly for convenience.
Right now, we do provide some helpers to convert to cupynumeric via
table.to_array()already, so this may be mainly a short utility to call these automatically.There may be a use/need to automatically cast columns into a common dtype, so that this may be more of an issue for legate-dataframe, where we would want to support
to_array(dtype=, null_value=).