[giordano]

The check of dynamic dimensions done when allocating a FixedSizeArray must be
cheap to inline: with loops in the Julia IR of checked_size_product it is not
inlined for 3 or more dimensions, and for 2 dimensions it makes the functions
allocating arrays and calling other functions, e.g. * of matrices, too
expensive to inline into their callers.

These tests require CheckedSizeProduct v1.0.1, where the reductions over the
tuple of dimensions are recursive rather than loops.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.39%. Comparing base (9452c7b) to head (b335bf6).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #230   +/-   ##
=======================================
  Coverage   94.39%   94.39%           
=======================================
  Files           8        8           
  Lines         321      321           
=======================================
  Hits          303      303           
  Misses         18       18           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[giordano]

All the test failures in https://github.com/JuliaArrays/FixedSizeArrays.jl/actions/runs/39948573136/job/120397168346?pr=230 (on commit ad7d713) show that the tests introduced here catch the issue that JuliaArrays/CheckedSizeProduct.jl#28 is resolving (#183)

[giordano]

Instead in https://github.com/JuliaArrays/FixedSizeArrays.jl/actions/runs/39952024839?pr=230 (on commit d5dbf6a) the same tests are passing everywhere except on Julia v1.10 (which doesn't read the [sources] section of Project.toml), to confirm that JuliaArrays/CheckedSizeProduct.jl#28 does indeed solve the issue (namely, increased cost of inlining of checked_dims, which prevents further optimisations)