Alternatives for IO

[klmr]

At the moment the io module exclusively uses R base IO plus xlsx.

There are two or three broad alternatives:

• data.table::fread. This is a thoroughly terrible idea and I strongly oppose it, because their code base and their API is a mess, and it’s been in development for years without a stable version. Performance is phenomenal, but we need correctness first, performance last.
• readr. Also still in development. Uses sane defaults (somewhat similar to what we’re already doing with stringsAsFactors), produces dplyr-compatible tbl_df without row names. Not as fast as fread but apparently still an order of magnitude faster than base R.
• rio. I honestly don’t know exactly what this is; it seems to be higher level, i.e. read data from different format using the same interface (same as io, but way more formats supported). Uses (or rather, will use in the future) readr under the hood.