ess-eval-region-or-function-or-paragraph-and-step fails to recognize comments between function head and body

[deepayan]

This is a minor but long-standing irritant possibly related to #1149. Try loading the following function definitions using C-c C-c:

## This one works fine

foo <- function(x, y, ...) { # This is a comment
    return(x + y)
}

## This one gives an error

bar <- function(x, y, ...) # This is a comment
{
    return(x + y)
}

## Selecting + C-c C-r works

The same behaviour is seen with these variants:

bar <- function(x, y, ...)
### x: first argument
### y: second argument
### ...: ignored
{
    return(x + y)
}

bar <- function(x, y, ...)
    ## x: first argument
    ## y: second argument
    ## ...: ignored
    return(x + y)

The only commonality that I can see that consistently triggers this bug is a comment between the formals and the body, but it also requires the function assignment. Both of the following work:

function(x, y, ...)
    ## This is a comment
{
    return(x + y)
}

(function(x, y, ...)
    ## This is a comment
{
    return(x + y)
})(4, 5)

This is not a showstopper by any means, but perhaps these reproducible examples will make it easier to fix.

[mmaechler]

This is actually a very very old bug (as opposed to 99% of the other open issues).
I've changed my habits because of this bug and am not using these comments anymore, something John Chambers very much on purpose used in S already ... and in some sense is even in the lisp tradition of "doc strings"...

A much younger -- but still about 10 years old -- bug is that there are quite a few other situations with larger functions where ESS is not able to find the beginning or end of the R function, e.g., (sometimes, I think not always) when you have an empty line in the function body; very often when your function defines local functions inside its body, ...