fix lu deprecation warnings on Julia nightly

[simeonschaub]

Ref JuliaLang/julia#40623

[oxinabox]

Should we put these in Compat.jl instead?

[simeonschaub]

I thought about that, but wasn't sure defining Compat.RowMaximum and Compat.NoPivot as just Val{true}/Val{false} was the right thing to do for Compat.jl, or whether it should define them as separate singletons and overload lu.

[oxinabox]

I don't think we are in any rush to fix this.
So lets open a PR on Compat.jl and see what the Compat.jl maintainers think?

[simeonschaub]

Ah, looking at the Julia PR again it seems like Val{true} only means RowMaximum for lu, while for qr Val{true} means ColumnNorm. This likely means we don't want to implement them as aliases for Val{true} in Compat, because lu(x, ColumnNorm()) would return lu(x, RowMaximum()) without any warning, which seems bad. I can open a PR defining these as singletons in Compat instead, but that would be independent of this PR, since we do actually need to overload lu(x, ::Val) on earlier Julia versions.

[oxinabox]

I can open a PR defining these as singletons in Compat instead, but that would be independent of this PR, since we do actually need to overload lu(x, ::Val) on earlier Julia versions.

let's do this

[oxinabox]

Do we need to call this LU_NoPivot
since we will want this to differ for QR?

[simeonschaub]

Yeah, makes sense

[codecov-commenter]

Codecov Report

Merging #434 (b4a36bf) into master (a1809f4) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #434   +/-   ##
=======================================
  Coverage   98.39%   98.39%           
=======================================
  Files          21       21           
  Lines        1989     1989           
=======================================
  Hits         1957     1957           
  Misses         32       32           

Impacted Files	Coverage Δ
src/rulesets/LinearAlgebra/factorization.jl	97.85% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a1809f4...b4a36bf. Read the comment docs.

[oxinabox]

merge when you are happy