Sam Isaacson

Those docs seem clear enough, they just postdate when the design here was created by quite a bit. (JumpProcesses' design goes back to before I started working on the library...

If we restrict this to a pure-jump integrator we shouldn't actually need continuous callbacks. So we can do that as a first pass.

Very weird, especially given that the allocations are staying constant. Note you should interpolate in calling `JumpProblem`, i.e. ```julia @btime JumpProblem($rn, $dprob, Direct()) ``` but that doesn't seem to change...

I'm not seeing this when directly building a `ConstantRateJump` based `JumpProblem`, but am seeing the same issue if you convert `rn` to a `JumpSystem` and build a system using the...

@ChrisRackauckas I think this is arising from https://github.com/SciML/ModelingToolkit.jl/blob/01aa16633be9fcb17a34c95e94c08ffa6d9d38a7/src/systems/jumps/jumpsystem.jl#L219 in MTK. Could this be due to repeatedly creating new `RuntimeGeneratedFunction`s?

This line https://github.com/SciML/ModelingToolkit.jl/blob/01aa16633be9fcb17a34c95e94c08ffa6d9d38a7/src/systems/jumps/jumpsystem.jl#L223 seems to be one of the culprits. I don't see any increase if I just call `generate_affect_function`, but I do see a consistent increase in running time...

Transferred to MTK as this isn't a Catalyst issue from what I can see.

This + the Sanft-Othmer project seem like a good GSOC for someone interested in data structures.

Generally the symbolic events are missing a lot of the normal event features. There is another issue about directionality for continuous events that @paulflang wanted for SBML model support, and...

This is now on master and will be in Catalyst 13