Brian Ward

Yep, the Stan repo and stanc3 would both also need changes to take advantage of it. For now I'm curious if @rok-cesnovar and I can make the PCH experience sufficiently...

The doc for `grad_hessian` requires that > The functor must implement `fvar operator()(const Eigen::Matrix&) ` which the model_base `log_prob`s currently do not. We'd need to seek upstream changes similar to...

`grad_hessian` as currently implemented doesn't seem to actually call either version of `hessian` directly, so it will probably be a bit more invasive of a change than just an include....

We must be talking past each other: - BridgeStan currently can call either the finite diff hessian or the "version 1" (`fvar`) function you mention above. No issues here, I...

The issue isn't that fewer models would work or anything, it's that at the moment we can only call overloads which exist in the `model_base_crtp` class, and `log_prob` does not...

Because the downstream usages do not directly call the constructor of the rng type directly and instead call the utility function that accepts a seed and chain ID, I don’t...

I think it's fine to want to use the same RNG for testing here as is used in Stan (though historically this wasn't done, and I could also see arguments...

In my opinion no - the real breaking change would actually be when we remove the flag and start using C++17-only constructs in the code. The utility of doing this...

@SteveBronder I think we will need another round of RCs for other changes, so we can! Because I opened this PR I can't actually click 'approve', so if you can...

Hi @alexandrebouchard - I'm one of the BridgeStan developers who just found this (now quite old) issue - do you still have any of the logs of this happening? It...