Suggestion: Kernighan and Ritchie style

[gvfarns]

The original indentation style pioneered by Kernighan and Ritchie in the book that defined the C language specified keeping open curly bracket on the same line for if, for, and other statements, but to put it on a line by itself for a function. The Linux kernel follows this style and so do I in my R code.

There is also a practical reason to use this formatting: vi/vim doesn't understand the beginnings of R functions unless the curly bracket is on a line by itself. As a result you can't highlight by function or easily move around by function in vim unless the curly bracket is by itself for a function. The same advantages are not present for other statements so putting it by itself everywhere is not necessary.

What I suggest is an option to enable the following formatting:

f <- function(x)
{
  if (x>0) {
    return(x^2)
  } else {
    return(-x^2)
  }
}

This is the formatting of the pure in heart. I think it would be great to be able to achieve this within the context of formatR. Thanks!

[yihui]

Styles in programming are not so much different from styles in dressing. In most cases, it does not really hurt if you use a different style. I understand people make different choices. I can certainly accommodate your style if the implementation is easy for me, but in this case, it is not easy unfortunately.

Personally I'm slightly against this style, because I do not want R code that is not supposed to be complete to be complete by chance (e.g. when you copy and run it), i.e.

{
  # function body
}

is a complete R expression when it is supposed to be the body of a function f:

f <- function()
{
  # function body
}

If you write

f <- function() {
  # function body
}

It is more unlikely that you copy a few lines of the code and it happens to be a complete expression.

No strong opinion here. Just a minor thought. Thanks for the suggestion anyway!