Chris, thanks for following up about the FFI topic. For my present project, Stan is what my collaborators require for and is the way to go because of the easy log likelihood calculation, so when time is available, I’d definitely like to contribute to the state of the differential equation solvers in Stan (which would also help me finish my projects and, alas, finally allow me to graduate). There is some additional ongoing discussion in this thread about figuring out differential equation development priorities in Stan here: Stan differential equation functionality development priorities.
Because the FFI situation was unknown, I had a growing interest in contributing to an auto-switching ODE solver, which would probably be the algorithm that would be most beneficial to my own projects. However, we did weigh some pros and cons of building solvers versus putting resources instead to forward an FFI. If the Stan devs do decide to put resources toward an FFI after further discussion, efforts towards building native solvers in Stan Math would be a less efficient and redundant goal.