Add ones for NCRing

[fingolfin]

Also simplify zeros implementation and add some tests (see also oscar-system/Oscar.jl#2231)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.13%. Comparing base (475eafc) to head (b03d06d).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1886      +/-   ##
==========================================
- Coverage   88.13%   88.13%   -0.01%     
==========================================
  Files         120      120              
  Lines       30286    30281       -5     
==========================================
- Hits        26694    26689       -5     
  Misses       3592     3592              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lgoettgens]

fill fills the array with aliases elements. We usually don't want this, but instead every zero or one entry should be it's own object

[fingolfin]

Hrm, I am torn on this.

We "usually don't want that" for our own matrices, but it is what ones and zeros does in Julia, e.g.

julia> M = ones(BigInt, 3, 3)
3×3 Matrix{BigInt}:
 1  1  1
 1  1  1
 1  1  1

julia> M[1,1] === M[1,2]
true

Of course this is not practical for actual use within OSCAR, but that's not what this is about.

Perhaps this is an argument to turn both of these functions into just error like error("please use zero_matrix instead") or so.

[thofma]

I don't consider it a problem that all elements are identical.

I don't really care

[lgoettgens]

We just had someone on slack that was trapped by the julia behavior.
Of course this is just a vote about how misleading the julia function is, not that we should do something else.

[fingolfin]

We had a discussion related to this during yesterday's triage and basically I now think that while "formally and legally" my new methods are "correct", they are footguns waiting to be triggered, so I now tend to agree that if we add ones it should be a modified duplicate of the existing zeros method, i.e. without aliasing.