Simplify show implementation

[lkdvos]

This PR simplifies a bit of the code used for defining show by leveraging the built-in Base functionality of replacing values for printing.
This shouldn't change the functionality, although it does change a bit of the exact alignment properties.
In particular, here, the zero-value print is used to determine things like alignment, and only afterwards is this replaced with a centered dot, while I think the previous implementation first replaced the entries with dots, and only then computed alignments.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 71.36%. Comparing base (1e54b85) to head (4216bd8).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #31      +/-   ##
==========================================
- Coverage   72.46%   71.36%   -1.11%     
==========================================
  Files           7        7              
  Lines         454      433      -21     
==========================================
- Hits          329      309      -20     
+ Misses        125      124       -1     

Flag	Coverage Δ
docs	37.44% <100.00%> (-2.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[mtfishman]

Does this work for higher dimensional arrays as well?

[lkdvos]

Yes, they recursively make views that they call print_matrix on. (don't mind the sprand, PR incoming for that)

julia> A = SparseArraysBase.sprand(Float64, (3,3,3), 0.2)
3×3×3 SparseArrayDOK{Float64, 3, typeof(SparseArraysBase.default_getunstoredindex)}:
[:, :, 1] =
  ⋅   0.884028   ⋅
  ⋅    ⋅         ⋅
  ⋅    ⋅         ⋅

[:, :, 2] =
  ⋅    ⋅         ⋅
  ⋅    ⋅         ⋅
  ⋅   0.859288   ⋅

[:, :, 3] =
  ⋅   0.716633  0.826796
  ⋅    ⋅         ⋅
  ⋅    ⋅         ⋅

[mtfishman]

Great, I think I tried to dig into Base.show to see how SparseArrays.jl did it but it wasn't clear to me if it generalized beyond matrices. Nice to see all of that code removed! Can you bump the version?

[lkdvos]

Honestly, I still think we should use the SparseArrayKit approach of simply printing all non-zero entries manually, because this implementation will still flood the terminal for N-D arrays that are large but sparse.
https://github.com/Jutho/SparseArrayKit.jl/blob/9a7c0a25b6ce6977c195c29ff119a96415da5685/src/sparsearray.jl#L174-L205
Possibly we could even support dynamically switching between the two, but that will be for a follow-up PR.

[mtfishman]

Honestly, I still think we should use the SparseArrayKit approach of simply printing all non-zero entries manually, because this implementation will still flood the terminal for N-D arrays that are large but sparse. https://github.com/Jutho/SparseArrayKit.jl/blob/9a7c0a25b6ce6977c195c29ff119a96415da5685/src/sparsearray.jl#L174-L205 Possibly we could even support dynamically switching between the two, but that will be for a follow-up PR.

I like the idea of switching over to just listing the stored values after a certain size (say once Julia would start to not show all of the values anyway).

[lkdvos]

This breaks a downstream BlockSparseArrays test, I didn't realize that the structure was being explicitly used there. Let me fix that first

[mtfishman]

This breaks a downstream BlockSparseArrays test, I didn't realize that the structure was being explicitly used there. Let me fix that first

Oh yeah, thanks.

[lkdvos]

I bumped to v0.3.0, but since #33 is also breaking, we might want to bundle these together

[lkdvos]

Can be merged after #33