Brian Groenke

My gripe isn't really the boilerplate, I agree that this is minor. It is more that I think this is a misuse of inheritance, and it's not clear to me...

Yes, all fine, the implementation of `update` that I had in mind: `update(m, [a], p -> p.optimizable)` would have to assume that the given vector matches the length of the...

> but did you mean Yes, that's pretty close to what I meant.

Sorry about that... hit the wrong button.

`IntervalSets` is a pretty lightweight package, I think (only three dependencies and a couple hundred lines of code). So I probably wouldn't worry about depending on it. For most users...

> Also thinking about this, doesn't Optim.jl take tuple bounds too? Or maybe two vectors. But I don't know if it handles intervals. It takes two vectors. So it doesn't...

Where is the code currently that actually handles the `bounds` field?

Ok, I'll take a look. It doesn't seem like it should be that hard... Regarding the different types issue, I think the open/closed distinction only makes sense for real numbers....

Component x field name can be duplicated. A simple example is a collection of objects of the same type. I have this problem at the moment, unfortunately. Could you elaborate...

Could I get some context on this? Why is this necessary? It seems to me that filtering is already possible by table operations on `Model`, and so-called "fixed" parameters can...