@inline StrideIndex

[chriselrod]

@ffreyer you can try this

[codecov]

Codecov Report

Merging #343 (c42c20b) into master (bf07fa9) will increase coverage by 0.16%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #343      +/-   ##
==========================================
+ Coverage   90.06%   90.23%   +0.16%     
==========================================
  Files           9        9              
  Lines        1319     1331      +12     
==========================================
+ Hits         1188     1201      +13     
+ Misses        131      130       -1     

Impacted Files	Coverage Δ
src/array_index.jl	100.00% <100.00%> (ø)
src/dimensions.jl	93.61% <0.00%> (+0.04%)	⬆️
src/stridelayout.jl	89.89% <0.00%> (+0.38%)	⬆️
src/ArrayInterface.jl	85.96% <0.00%> (+0.64%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[ffreyer]

Doesn't seem to help

julia> @benchmark reflectorApply!($x, $τ, $y)
BenchmarkTools.Trial: 10000 samples with 1 evaluation.
 Range (min … max):  46.028 μs … 210.703 μs  ┊ GC (min … max): 0.00% … 0.00%
 Time  (median):     49.761 μs               ┊ GC (median):    0.00%
 Time  (mean ± σ):   52.597 μs ±  11.745 μs  ┊ GC (mean ± σ):  0.00% ± 0.00%

  ▇▇███▇▅▃▁    ▁▂▃▃▃▃▂          ▂▂▁                            ▃
  ███████████▆▅███████▇▅▆▆▅▄█▇█████▇▅▅▅▅▅▅▆▇▇▇█▆▆▆▃▄▃▁▁▃▄▆▃▆▆▅ █
  46 μs         Histogram: log(frequency) by time       105 μs <

 Memory estimate: 0 bytes, allocs estimate: 0.

(@v1.7) pkg> st ArrayInterface
      Status `~/.julia/environments/v1.7/Project.toml`
  [4fba245c] ArrayInterface v6.0.22 `https://github.com/JuliaArrays/ArrayInterface.jl.git#inlinestrideindex`

[Tokazama]

We might need a more aggressive fix for for previous versions

function map_tuple_type(f::F, ::Type{T}) where {F,T<:Tuple}
    if @generated
        t = Expr(:tuple)
        for i in 1:fieldcount(T)
            push!(t.args, :(f($(fieldtype(T, i)))))
        end
        Expr(:block, Expr(:meta, :inline), t)
    else
        Tuple(f(fieldtype(T, i)) for i in 1:fieldcount(T))
    end
end

[chriselrod]

We might need a more aggressive fix for for previous versions

You're welcome to take over this PR, or we can merge it and you follow up with another?

I was waiting on integration tests to merge, but this PR seems like it should be harmless w/ respect to semantics.

[Tokazama]

We might need a more aggressive fix for for previous versions

You're welcome to take over this PR, or we can merge it and you follow up with another?

I was waiting on integration tests to merge, but this PR seems like it should be harmless w/ respect to semantics.

Sounds good. I put in the old generated methods. I'll wait for some tests to pass before merging but your probably right that we don't need to wait for all of the integration tests.