PgDatabaseMetaData.getIndexInfo() cast operands to smallint

[jsyrjala]

It is possible to break method PgDatabaseMetaData.getIndexInfo() by adding certain custom operators.
This PR casts operands in a query so that PgDatabaseMetaData.getIndexInfo() works in presence
of such operator config.

Creating this operator triggers this issue:

CREATE OR REPLACE FUNCTION f6(numeric, integer) returns integer as 'BEGIN return $1::integer & $2;END;' language plpgsql immutable;
CREATE OPERATOR & (LEFTARG = numeric, RIGHTARG = integer, PROCEDURE = f6);

Fixes #2241

All Submissions:

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same update/change?

Changes to Existing Features:

• Does this break existing behaviour? If so please explain.
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you written new tests for your core changes, as applicable?
• Have you successfully run tests with your changes locally?

[davecramer]

adding an operator like this to the test cases would be useful

[jsyrjala]

What would be the right place to add the test case? There are several DatabaseMetaDataTest classes. Only one of those calling getIndexInfo() is jdbc2.

./java/org/postgresql/test/jdbc42/DatabaseMetaDataTest.java
./java/org/postgresql/test/jdbc4/DatabaseMetaDataTest.java
./java/org/postgresql/test/jdbc3/DatabaseMetaDataTest.java
./java/org/postgresql/test/jdbc2/DatabaseMetaDataTest.java

[davecramer]

it's pretty fundamental so jdbc2 is fine.

[jsyrjala]

Added the operator creation to ./java/org/postgresql/test/jdbc2/DatabaseMetaDataTest.java. Having the operator makes a couple of tests in DatabaseMetaDataTest fail with code from master branch. Doing the casts in getIndexInfo() makes the tests pass again.

[davecramer]

Thx!

[jsyrjala]

I guess this PR is now directed against upcoming 43.x development version? Would it possible get this fix also to 42.2.x?

[davecramer]

yes, of course

[jsyrjala]

Should I do an another PR with the same content against release/42.2 branch?

[davecramer]

Yes, please