[General] JOSS paper review

[tmigot]

Thank you for the paper. I found it strong and well written overall, and the package looks like a very valuable contribution to the Julia optimal-control ecosystem. I have a few minor comments/suggestions that could help improve clarity and consistency.

• Make the styling/linking of .jl packages uniform.
  Some packages appear with links/citations while others do not, e.g., DifferentiationInterface.jl and ForwardDiff.jl. It would be nice to make the formatting consistent and, when possible, hyperlink or cite all mentioned .jl packages.

• Add a citation for MINPACK.
  MINPACK.jl/MINPACK is mentioned in the indirect-methods paragraph, but I did not see a corresponding citation.

• Check the companion example citation/version.
  The illustrative example seems to refer to the companion repository:
  https://github.com/control-toolbox/joss-oc-example
  Please check that the citation in the paper points to the correct archived version. I believe it may need to be updated from v1.0.0 to v1.0.1.

• Add or repeat the Ipopt citation in the “Modelers and solvers” paragraph.
  In the paragraph listing NLP solvers, it may be useful to cite Ipopt directly there as well, not only NLPModelsIpopt.jl.

• Clarify the exact class of optimal control problems currently supported.
  The title says “optimal control problems with ODEs,” which is clear, but the paper would benefit from a compact summary of the mathematical problem class supported by the package. For example, could the authors clarify whether the current implementation supports:

  • fixed and free final time;
  • endpoint constraints;
  • path constraints;
  • box constraints on states and controls;
  • Mayer, Lagrange, and/or Bolza costs;
  • autonomous and non-autonomous dynamics;
  • multiphase problems;
  • DAEs or only explicit ODEs;
  • hybrid/nonsmooth systems.

  Some of these points may be obvious to optimal-control specialists, but for a broader JOSS readership it would be helpful to make the supported scope explicit.