[Rd] New pipe operator

[Gabor Grothendieck]

Duncan Murdoch:
> I agree it's all about call expressions, but they aren't all being
> treated equally:
>
> x |> f(...)
>
> expands to f(x, ...), while
>
> x |> `function`(...)
>
> expands to `function`(...)(x).  This is an exception to the rule for

Yes, this is the problem.  It is trying to handle two different sorts of right
hand sides, calls and functions, using only syntax level operations and
it really needs to either make use of deeper information or have some
method that is available at the syntax level for identifying whether the
right hand side is a call or function.  In the latter case having two
operators would be one way to do it.

  f <- \(x) x + 1
  x |> f()  # call
  x |:> f  # function
  x |:> \(x) x + 1  # function

In the other case where deeper information is used there would only be one
operator and it would handle all cases but would use more than just syntax
level knowledge.

R solved these sorts of problems long ago using S3 and other object oriented
systems which dispatch different methods based on what the right hand side is.
The attempt to avoid using the existing or equivalent mechanisms seems to have
led to this problem.