Bump ImplicitDifferentiation bound + run CompatHelper on subpackages

[gdalle]

I just realized your ControlSystemsBase was stuck on an old version of ImplicitDifferentiation (0.4 when we're at 0.7), so I checked and your CompatHelper workflow doesn't see subdirectories. This PR should fix it, and CompatHelper should open new PRs itself soon to upgrade ImplicitDifferentiation

See also:

• https://juliaregistries.github.io/CompatHelper.jl/dev/options/#Packages-in-subdirectories

[baggepinnen]

Thanks for the PR! It seems the extension needs an update to be compatible with the latest version of DI. Do you recognize this error https://github.com/JuliaControl/ControlSystems.jl/actions/runs/14397751088/job/40385323705?pr=980#step:5:61 ? It seems some internal function has gotten a different type signature?

[gdalle]

Oh, that's probably because the new version of ImplicitDifferentiation only accepts vectors instead of general arrays. It's easier to code that way, but if it causes real problems downstream I may reconsider it

[baggepinnen]

What's the intended workflow for handling matrices? vec -> reshape?

[gdalle]

What's the intended workflow for handling matrices? vec -> reshape?

Yeah exactly, either I do it myself in ImplicitDifferentiation or every user does it. Inside ID it's easier to work with vectors because we need to solve a linear system, which is more naturally formulated with matrices. But you're not the first user for whom this is a problem, so I'll see what I can do

[gdalle]

JuliaDecisionFocusedLearning/ImplicitDifferentiation.jl#167

[gdalle]

@baggepinnen could you please approve the workflows? The last commit picks the version of ImplicitDifferentiation where arrays are supported again

[gdalle]

Oh right, I see what's happening here. Your first ImplicitFunction did work with a byproduct z so I didn't bother to check the rest, but not all of them do actually. In the new version of ImplicitDifferentiation, byproducts are mandatory. Fixing it right now.
EDIT: actually, the problem was that hinfnorm returned a scalar instead of an array.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.42%. Comparing base (5ab74cf) to head (96e6a4d).
Report is 6 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #980      +/-   ##
==========================================
+ Coverage   92.41%   92.42%   +0.01%     
==========================================
  Files          41       41              
  Lines        4981     4990       +9     
==========================================
+ Hits         4603     4612       +9     
  Misses        378      378              

☔ View full report in Codecov by Sentry.
📢 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.

[gdalle]

All good now I think

[baggepinnen]

Thanks for all your fixes :)

[gdalle]

Turns out the ImplicitDifferentiation ones were not needed after all😅 but it was a good excuse.
As a side note, since you're mostly working with scalars, this code would probably benefit tremendously from StaticArrays