luchr
Wrap ROCK/RKC solvers?
You can find ROCK2 and ROCK4 from Hairer's site as well as mildly-stiff integrators. I think this would be a good home for them.
(also brainstorming, would you include the DDE codes RETARD and RADAR5 from that same page?)
For RETARD the answer is simply: No. RETARD needs at compile time common blocks with data-structures that depend on the number of "back steps". So the number of (maximal) supported back steps and dense components must be known at compile time, which is not state-of-the-art. The amount of rewriting RETARD to pass-through an additional workspace pointer on the stack (instead of using this common blocks) is in my view to large. RADAR: inclusion/interface: might be
Bumping the question for the ROCK solvers. Everything else that Hairer has for standard ODE/DDE (which benchmarks well, i.e. exclude extrapolation) has some native Julia alternative. But those ROCK2 and ROCK4 methods are peculiar and I would love to be able to give them a try and benchmark them against the rest of the methods.