Hi folks,
I've submitted two performance related patches for schema matching (SQOOP-1968
and SQOOP-1969). While investigating the issue, I noticed that every
`getMatchingData(fields)` call will check if any field represents a NULL. It
tried to set a field it to null, if it equals ("NULL", "null", "'null'", "" or
null). It looks like a special handling for database exported CSV null values.
I feel it should be moved to CSVIntermediateDataFormat. So I've created
SQOOP-1988.
I'll be very appreciated to everyone who has participle this discussion in
mailing list or on JIRA (SQOOP-1988 and SQOOP-1811).
--Stanley (Qian) Xu
