Codecov Report

Merging #318 (2945d61) into master (08dc371) will decrease coverage by 1.29%.
The diff coverage is 50.00%.

@@            Coverage Diff             @@
##           master     #318      +/-   ##
==========================================
- Coverage   96.45%   95.16%   -1.30%     
==========================================
  Files           5        5              
  Lines         451      434      -17     
==========================================
- Hits          435      413      -22     
- Misses         16       21       +5     

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

Looking at the example from #316,

julia> Base.@kwdef struct MyIdOffsetRange
                  values::UnitRange
                  indices::UnitRange
              end
MyIdOffsetRange

julia> x = MyIdOffsetRange(values=5:8, indices=5:8)
MyIdOffsetRange(5:8, 5:8)

julia> x.values
5:8

julia> x.values === x
false

In this example, the returned results correspond exactly to the displayed ones, as both directly read the fields of the struct. However, for an IdOffsetRange, we obtain another IdOffsetRange, which isn't what was displayed.

julia> id
OffsetArrays.IdOffsetRange(values=5:8, indices=6:9)

julia> id.values
OffsetArrays.IdOffsetRange(values=5:8, indices=6:9)

julia> id.values === id
true

This isn't an objection to the PR, but I wonder if we need to be consistent here?

The consistency choice I made here was for

id.values === values(id)

values is part of the API already.

If you really wanted a UnitRange, this could work:

UnitRange(id.values)

For that reason, I added it to the documentation.