Documentation: more explicit that j can be logical (when with=FALSE)

[Henrik-P]

I stumbled over one of the highly informative error messages in data.table:

When with=FALSE, j-argument should be of type logical/character/integer indicating the columns to select.

However, the fact that j can be of type logical is not explicitly described in ?data.table.

From with:

When with=FALSE j is a character vector of column names, a numeric vector of column positions to select or of the form startcol:endcol

From j:

When with=FALSE, j can only be a vector of column names or positions to select

[franknarf1]

Fyi, if you're writing new code and using version 1.9.8+, you might as well skip the with=FALSE bit:

library(data.table)
DT = data.table(a = 4, b = 5)
DT[, 2]
DT[, 2, with=FALSE] # same

[Henrik-P]

Thank you for the heads-up! Yes, I was aware of that. It was actually after reading the NEWS "Changes in v1.9.8", section 3 about with= that I started to play around with simple (and sometimes veery stupid...) examples. I then stumbled over the error message, which led me to think about logical j and how it was described in the docs. Cheers.

[MichaelChirico]

Funnily, I don't see the fact that j can be logical in the documentation for [.data.frame either...