>> - "make check" fails 16 of 118 tests for me with this patch applied.
> Most of them are caused by additional NOTICE messages or unexpected
> additional rules in the rewriter regression tests. I don't see anything
> critical here.
Possible; in that case you should patch the expected regression output
appropriately. But I seem to remember thinking that \d was producing
the wrong column list on one of the system catalogs you modified, so
you might want to double check that part... maybe I'm all wet.
...Robert