5

Are there any significant [1] reasons not to export functions with names of the form .fnname in R packages?

I understand that the main use of dot-prefixed variables is to denote a variable as being hidden when searching through an environment using functions like ls, and to signify that fields in an object or list should be treated as being private like the S4 .Data field. 

test_env           <- new.env(parent = emptyenv())
test_env $ .hidden <- 10

ls(test_env)
# character(0)

ls(test_env, all.names = TRUE)
# ".hidden"

As far as I can tell the convention is enforced at the level of a few utility environment-searching functions but not throughout the R language as a whole.

Does dot-prefixing a variable alter the behaviour of core language features like lexical scoping or the various object systems, and if so would this make exporting dot-prefixed functions inappropriate or dangerous?

Thanks for any help or info you can offer.

Note:

[1] By significant I mean non-stylistic; the library in question uses special prefixes and suffixes to flag aspects of a function's type signature, somewhat like plyr's special naming convention. It would normally be non-semantic to abuse dot-prefixed variables, but in this case it is worth the compounded meaning.

Róisín Grannell
  • 2,048
  • 1
  • 19
  • 31
  • 6
    I think , It is just a convention and don't alter the core behavior in any manner. – agstudy Nov 05 '14 at 09:47
  • 1
    I've just skimmed through the R language definition and couldn't find a mention of this. Thus, I'd agree with @agstudy. – Roland Nov 05 '14 at 09:56

0 Answers0