It is unclear today as to what is the optimal place where the type mappings described in the
SQOOP-515 between the From and To can happen.
For example in case of HIVE, if the IDF FixedPoint size attribute is given, the destination schema column type might be dependent on the value of this size is in the corresponding From schema.
For example in Hive.
if size < 1 then tinyint
if size < 2 then smallint
if size < 4 then int
Read more: https://issues.apache.org/jira/secure/attachment/12589331/Sqoop2Datatypes.pdf
To summarize the question s how does the ToInitializer get access to the schema from the "FROM"