James Mitchell

I seem to remember that `Random(Integers)` always return a value in `[-12, 12]` or something similar, > exaggeratedly concentrated around 0 sounds about right :)

Here's another example of an integer matrix with non-integer entries: ~~~ gap> Matrix(Integers, [[1, 1], [2, E(8)]]); gap> IsInt(E(8)); false ~~~ Again maybe this is the intended behaviour, but this...

I agree with @ThomasBreuer, I think the "correct" behaviour would be for `Inverse` and `Matrix(Integers, [[1, 1], [2, E(8)]])` to return `fail` or give an error, and that there should...

Seems completely reasonable to me @hulpke, it's what we did in `Semigroups` and it's the test failures there that indicated this issue to me when converting from our own matrices...

Thanks for raising this, and apologies for breaking the CI @fingolfin, any chance you can link me to the log showing the failure?

> I'm unsure why the package installs such methods (that apply to semirings which are fields) and doesn't just use the matrices from the library. When we did this back...

> @james-d-mitchell to be clear I don't think it'd be fair to see that "you" broke the CI. If at all, most like it's my patch for semigroups that's the...

> In fact, only removing the artificial upranking in lines 304 and 310 of that file make the error disappear. Looks like this only makes the error disappear in the...

Thanks @hulpke, I think a combination of 3 and 4 is probably the best way forwards. It'd be better for semigroups to use the better code for matrices that GAP...

Thanks for your comments, @fingolfin and @ThomasBreuer. I agree this is probably too big a change to make all at once, perhaps the right thing to do is to merge...