Re: [HACKERS] GSoC 2017: Foreign Key Arrays

On Sat, Mar 13, 2021 at 12:08:26PM +0200, Mark Rofail wrote:
> > I guess the failure is in the first patch, but isn't caught by test cases
> > until the 2nd patch
>
> This actually uncovered that I didn't add tests for the new operators in
> `gin.sql`, I have done so now, hopefully, that will help us uncover the
> problem. Can you post the regression results for the v8 patch attached
> below?

> +++ b/src/test/regress/expected/gin.out
> +select count(*) from gin_test_tbl where i @>> 1;
> + count
> +-------
> + 3
> +(1 row)
> +
> +select count(*) from gin_test_tbl where i @>> 999;
> + count
> +-------
> + 0
> +(1 row)

I think the expected results are actually wrong, with the linux CI failing but
actually returning the right results.

I have a fix which passes (modified) tests, although I doubt it's complete.

I think this patch will need additional review - more than I can give.

--
Justin