Peter Eisentraut <peter_e@gmx.net> writes:
> A regression test for EXPLAIN, however, should primarily check whether the
> output format is stable. We are planning to offer this as a public interface,
> after all. You could use faked up statistics and all but one or two plan
> types turned off, and then the results should be pretty stable.
You'd be surprised :-(. We've found in the past that queries in the
regression tests get different plans across different platforms just
because of alignment-rule differences (leading to different numbers
of rows per page, etc etc). I think that test cases could be chosen
to be relatively stable points in the plan space, but it's hopeless
to imagine that the low-order digits of cost estimates will be the
same across all platforms.
regards, tom lane