Fredrik Johansson

I agree this needs to be done. Who's willing to make a pull request? I think I know how to patch the makefile, but the CMake build has to be...

> and properly used by downstream That is one massive assumption :-) > What about tests - do tests use the installed location? No.

I don't think dicts are the issue, or at least they shouldn't be (if they are, there's probably some low-hanging optimization to be done). But yes, you could use Arb....

I don't know what would be causing this. Could you try to track down the point of failure in the code, and the values of the variables in the t-gauss_period_minpoly...

If I'm reading this correctly, the the problem occurs when q = 0. In this case, arb_fmpz_poly_gauss_period_minpoly should set pol to the zero polynomial, which means the if (!fmpz_poly_is_zero(pol)) branch...

I tried compiling the test with GCC-7; no error.

Thanks. I agree that a compiler bug is plausible. It's also possible that the code contains inadvertent undefined behavior that causes problems only with the optimizations in recent GCC:s. It...

There is probably a less invasive way to make the test pass. Some possibilities: 1. Insert some non-inline function calls to prevent the compiler from simplifying. For example q =...

Thanks for getting to the bottom of this. Great work!

With a fix now in flint trunk, I guess we can just add a workaround to the test code so that it passes with flint-2.5?