Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
0.7.0
-
None
-
None
-
Reviewed
Description
In case of two relations being loaded using different loader, joined, grouped and projected, pig gets confused in trying to find appropriate loader for the requested cast. Consider the following script :-
A = LOAD 'data1' USING PigStorage() AS (s, m, l);
B = FOREACH A GENERATE s#'k1' as v1, m#'k2' as v2, l#'k3' as v3;
C = FOREACH B GENERATE v1, (v2 == 'v2' ? 1L : 0L) as v2:long, (v3 == 'v3' ? 1 :0) as v3:int;
D = LOAD 'data2' USING TextLoader() AS (a);
E = JOIN C BY v1, D BY a USING 'replicated';
F = GROUP E BY (v1, a);
G = FOREACH F GENERATE (chararray)group.v1, group.a;
dump G;
This throws the error, stack trace of which is in the next comment