Switch to _copy_oftype

[rikhuijzer]

Follow up on JuliaArrays/FillArrays.jl#172. Still part of trying to get LazyArrays to work on Julia 1.8-beta1: JuliaArrays/LazyArrays.jl#205.

With this change, the tests still pass on earlier Julia versions. Julia 1.8-beta1 fails which I've narrowed down to JuliaLang/LinearAlgebra.jl#918. Also, there is a problem with a missing eltype for a certain signature which I haven't been able to narrow down yet.

[dlfivefifty]

We need to make the compat entry require the version of FillArrays that introduces _copy_oftype

[codecov]

Codecov Report

Merging #88 (57d6d5a) into master (11aa76e) will decrease coverage by 0.54%.
The diff coverage is 84.21%.

@@            Coverage Diff             @@
##           master      JuliaLang/julia#88      +/-   ##
==========================================
- Coverage   85.38%   84.84%   -0.55%     
==========================================
  Files          10       10              
  Lines        1553     1603      +50     
==========================================
+ Hits         1326     1360      +34     
- Misses        227      243      +16     

Impacted Files	Coverage Δ
src/mul.jl	82.60% <ø> (-1.21%)	⬇️
src/ArrayLayouts.jl	81.98% <42.85%> (-3.82%)	⬇️
src/factorizations.jl	90.44% <89.47%> (+0.59%)	⬆️
src/diagonal.jl	84.31% <100.00%> (ø)
src/muladd.jl	74.48% <100.00%> (-1.18%)	⬇️
src/ldiv.jl	77.00% <0.00%> (-1.90%)	⬇️
src/lmul.jl	85.45% <0.00%> (-0.52%)	⬇️
... and 3 more

---

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 11aa76e...57d6d5a. Read the comment docs.

[dlfivefifty]

There seems to be some changes to elsize. We'll have to look into this (it might involve creating our own _elsize)

[rikhuijzer]

There seems to be some changes to elsize. We'll have to look into this (it might involve creating our own _elsize)

I just came across JuliaLang/julia#44674 so it seems you luckily have a solution for that now.

Based on the reply that I got on JuliaLang/LinearAlgebra.jl#918, I've now changed the equality in all(mul(Y',X) .=== Y'X) to isapprox in 47611bc.

The tests now still fail with elsize related errors on Julia 1.8-beta1. Oddly enough, the tests do pass on an 1.9 nightly that I also tested. I still haven't been able to figure out what PR in Julia base caused this change.

[dlfivefifty]

I have the tests now working, about to push

[rikhuijzer]

For me this is good to merge like this 👍 I cannot press the "approve review" button because I'm the author of the PR.