Tomás Chor

> > But I'm assuming we want to actually make this work before merging, correct? > > Of course we do. I need your help. I was a bit confused...

> > > > But I'm assuming we want to actually make this work before merging, correct? > > > > > > > > > Of course we do....

> I extended the implementation to try to support your use case @tomchor. My original implementation, which was meant just to show the source code changes required, only supported `backend...

> > This is a separate discussion, but I think ideally in the future we'd have situations where we'd be able to do something like > > I don't completely...

> > > > This is a separate discussion, but I think ideally in the future we'd have situations where we'd be able to do something like > > >...

That's cool! Can we make `getindex()` behave like that for `Field`s by default? `view(u[10], 1, 2:4, 2)` is not super intuitive.

> > That's cool! Can we make `getindex()` behave like that for `Field`s by default? `view(u[10], 1, 2:4, 2)` is not super intuitive. > > I am hesitant to do...

> > > > That's cool! Can we make `getindex()` behave like that for `Field`s by default? `view(u[10], 1, 2:4, 2)` is not super intuitive. > > > > >...

I vote to for an `add_` prefix in the function name. I don't think it's ugly and it clarifies things a fair bit. Also, rather than creating new structs for...

> > Also, rather than creating new structs for the format, why not use the actual writer constructors in the user interface? > > In a future where the "writer"...