Adam Jermyn

I think that's a great way to consolidate. More detailed thoughts: > Each set on inputs (star_job, controls, pgstar) should just be moved to their own s% struct (star_job is...

I think a fair amount gets re-evaluated each mdot loop. It's mdot and eps_mdot and element diffusion and rotation and premixing/predictive mixing, I think. > If we assume that everything...

> But I'm not sure about s% step% iter, for a lot of stuff I think it would defeat the purpose to have things like 's% step% iter% lnR(k)' Remember...

> I know that, was bringing it up because I did not see the benefit of having that additional sublayer of information. Thought the original idea from Rob was regarding...

Fully agreed that this needs broad discussion, and that it shouldn't happen before the release (except, as Rob says, in a branch). It'll be a decent amount of work too,...

As summary of where we are for further discussion, the current proposal I believe is to define five sub-structs that live inside the star pointer. These would divide up the...

Something like this? (Happy to iterate on this...)

The `implicit_mdot_loop` is only a loop if we set `mdot` implicitly. Rotation is currently not implicit but could become so (I think @orlox is/will be working on this). I just...

Ok. So then to summarize where we are: The proposed division is: 1. Things set by inlists go in s%star_job, s%controls, s%pgstar. 2. The minimum state needed to define the...

We sometimes need to compute finite differences in time between the end of `adjust_mass` and the current value inside the Newton iteration loop. This can be done by storing a...