CompatHelper: bump compat for "SpecialFunctions" to "2"

[github-actions]

This pull request changes the compat entry for the SpecialFunctions package from 0.8, 0.9, 0.10, 1.0 to 0.8, 0.9, 0.10, 1.0, 2.

This keeps the compat entries for earlier versions.

Note: I have not tested your package with this new compat entry. It is your responsibility to make sure that your package tests pass before you merge this pull request.

[mateuszbaran]

Blocked by (at least) JuliaStats/Distributions.jl#1434

[stevengj]

It's not blocked, as commented here.

[devmotion]

Blocked by (at least) JuliaStats/Distributions.jl#1434

The latest releases of ForwardDiff and Distributions are compatible with SpecialFunctions 2.0.

It's not blocked, as commented here.

On Julia >= 1.7 tests will error if the latest versions of the dependencies that is declared to be compatible can't be installed (see, e.g., https://github.com/JuliaManifolds/Manifolds.jl/runs/4512695265?check_suite_focus=true#step:5:30 and https://github.com/JuliaStats/Distributions.jl/runs/4500903661?check_suite_focus=true#step:6:71). I.e., you have to set force_latest_compatible_version=false manually (https://github.com/julia-actions/julia-runtest/blob/a5f2948fcb2fe71a428a722c286a48efd45bc610/action.yml#L18) in the Github action. Otherwise tests will fail, or rather not even start.

[mateuszbaran]

Blocked by (at least) JuliaStats/Distributions.jl#1434

The latest releases of ForwardDiff and Distributions are compatible with SpecialFunctions 2.0.

Right, though now ReverseDiff is still incompatible with SpecialFunctions 2.0: JuliaDiff/ReverseDiff.jl#188 .

[devmotion]

I'm on it, I'll merge and tag a new release when tests pass: JuliaDiff/ReverseDiff.jl#188 It took a bit more time since I had to fix some unrelated test errors first.

[devmotion]

JuliaRegistries/General#50686

[mateuszbaran]

Thanks for the update! The next one, it seems, is Zygote.jl.

[devmotion]

Zygote already merged the SpecialFunctions update without testing it a while ago: FluxML/Zygote.jl#1122

[mateuszbaran]

Yes but they haven't tagged a new version since then.

[devmotion]

Another main missing piece (not sure if relevant for Manifolds) is CUDA. However, a new release with SpecialFunctions 2 support is already prepared and will be available soon (hopefully).

[devmotion]

I requested a new release (I'm not familiar with Zygote PRs and releases, and hence let more knowledgeable people make a release): FluxML/Zygote.jl#1137

[devmotion]

A new Zygote release will be available soon: JuliaRegistries/General#51051

[codecov]

Codecov Report

Merging #444 (6ed63bf) into master (d99df9d) will decrease coverage by 0.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #444      +/-   ##
==========================================
- Coverage   97.99%   97.97%   -0.02%     
==========================================
  Files          82       82              
  Lines        6529     6531       +2     
==========================================
+ Hits         6398     6399       +1     
- Misses        131      132       +1     

Impacted Files	Coverage Δ
src/manifolds/Rotations.jl	95.58% <0.00%> (-0.41%)	⬇️
src/manifolds/Sphere.jl	100.00% <0.00%> (ø)
src/manifolds/Euclidean.jl	100.00% <0.00%> (ø)
src/manifolds/ProjectiveSpace.jl	100.00% <0.00%> (ø)
src/manifolds/ProductManifold.jl	94.13% <0.00%> (+0.01%)	⬆️

---

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 d99df9d...6ed63bf. Read the comment docs.

[kellertuer]

Looks good in general, can we shortly check where we lose the one line code coverage? Is that just a misclassification?

[mateuszbaran]

This lost coverage may either be due to a minor change in Distributions.jl or just a random error in coverage calculation. I'd just ignore it.