Correct mutating op fallbacks

[lgoettgens]

Base.inv and AbstractAlgebra.inv differ in handling Integers. The base version returns floats, while the AA version errors if it doesn't get a unit.

Similarly, Base.div and AbstractAlgebra.div differ. For this function, we need to either change the fallback of div! from AbstractAlgebra.div to Base.div, or change the docstring in a similar way as for inv!. Which of the two do we want here?

(In the any case, #1816 needs to be adjusted to be consistent.)

[codecov]

Codecov Report

Attention: Patch coverage is 57.14286% with 3 lines in your changes missing coverage. Please review.

Project coverage is 88.03%. Comparing base (11bbd58) to head (d7eb617).
Report is 5 commits behind head on master.

Files with missing lines	Patch %	Lines
src/fundamental_interface.jl	50.00%	2 Missing ⚠️
src/Fields.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1823   +/-   ##
=======================================
  Coverage   88.03%   88.03%           
=======================================
  Files         119      119           
  Lines       29993    29985    -8     
=======================================
- Hits        26403    26397    -6     
+ Misses       3590     3588    -2     

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

[thofma]

Complicated. They need to fall back to the Base version, since this is what one should overload. But they should not fallback to the Base version for e.g. BigInt.

Edit: Nevermind. They should fallback to the AA.inv an AA.div version.

[thofma]

I am not sure a docstring inside AbstractAlgebra needs to use the AbstractAlgebra prefix.

[lgoettgens]

I am not sure a docstring inside AbstractAlgebra needs to use the AbstractAlgebra prefix.

Including this docstring into Oscar should prefix. I'll extend this a bit and maybe explain the difference there

[fingolfin]

It can be very surprising (esp. for a Julia beginner) that AbstractAlgebra.inv != Base.inv so I think it is good to prefix it here explicitly to raise the awareness. Even more so for when this appears in the Oscar manual.

Should the same be done for div?

[fingolfin]

Ah I see you wrote that as a question. So yeah, I think div! should fallback to AA.div! (not Base.div!) and then its docstring should also be adjusted.

[lgoettgens]

I now adapted the docstrings of inv! and div!.
Furthermore, I encountered some methods that are attached to the wrong function. This shouldn't change any behavior due to the delegation from AA.* to Base.*, but makes it more consistent

[lgoettgens]

redundant due to

AbstractAlgebra.jl/src/algorithms/GenericFunctions.jl

Lines 77 to 79 in c70d3b6

	function Base.div(a::T, b::T) where T <: RingElem
	return divrem(a, b)[1]
	end