I don’t have a hard line, really — I’m not dogmatic by nature, which is probably why I generally dislike Tidy!

I mentioned lubridate and stringr, and those remain occasionally useful, but really only for one or two functions from each, that just have higher level interfaces and save me some time in interactive work. When I’m writing my own utilities and such, I still avoid them.

The big elephant in the Tidy room is ggplot2. Which is weird since it predates Tidy, and actually shares little in the way of semantics. I dislike it too, for some of the same philosophical choices. (And that’s despite reading Cleveland’s book on a “grammar of graphics” years ago and generally finding it a good idea!) The thing is, most people write libraries built on ggplot, not lattice or base R — so I do pick it up occasionally. I admit, I barely even understand how it does anything. Using “aes” for things that literally aren’t esthetic considerations, and the “+” symbology, irks me every time.

I continue to use R, and teach it. But there may come a day when the ecosystem is so thoroughly tidified that I’ll be forced to switch to Julia.