Re: Regression tests vs existing users in an installation

On Sat, Jul 16, 2016 at 7:13 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> We've talked before about how the regression tests should be circumspect
> about what role names they create/drop, so as to avoid possibly blowing
> up an installation's existing users during "make installcheck". In
> particular I believe there was consensus that such names should begin
> with, or at least include, "regress". I got around today to instrumenting
> CreateRole to see what names we were actually creating, and was quite
> depressed as to how thoroughly that guideline is being ignored (see
> attached).

Thanks for doing this.

> A more aggressive answer would be to decide we don't need these test cases
> at all and drop them. An advantage of that is that then we could
> configure some buildfarm animal to fail the next time somebody ignores
> the "test role names should contain 'regress'" rule.

I am -1 for dropping the tests. We could just have a CFLAGS that adds
an elog(ERROR) in CreateRole and checks that the created role has a
wanted prefix, or have a plugin that uses the utility hook to do this
filtering.
--
Michael