Currently, correlations are supported in the parser and query block model analyzer, but cause the translator to crash. In addition, having correlations in the algebra graphs will cause problems in other modules like the provenance rewriter and the optimizer.
first implement support in the translator
then add support in the provenance rewriter
then extend the optimizer to support this
finally support it in the SQL serializer
Considerations:
can we get around treating correlated attributes as special in certain places?
what data structures best to use to support correlations in the algebra graphs
need fast navigation to the operator from which the correlated attribute originates from
need convenience functions that check whether correlations exist in a subtree
Currently, correlations are supported in the parser and query block model analyzer, but cause the translator to crash. In addition, having correlations in the algebra graphs will cause problems in other modules like the provenance rewriter and the optimizer.
Considerations: