[Rd] A few suggestions and perspectives from a PhD student

[Lionel Henry]

> Third, there's a lot of nonstandard evaluation going on in all these
> packages. Using them inside your own functions requires serious attention
> (eg the difference between aes() and aes_() in ggplot2). Actually, even
> though I definitely see the merits of these packages in data analysis, the
> tidyverse feels like a (clean and powerful) macro language on top of R.

That is going to change as we have put a lot of effort into learning
how to deal with capturing functions. See the tidyeval framework which
will enable full and flexible programmability of tidyverse grammars.

That said I agree that data analysis and package programming often
require different sets of tools.

Lionel