[Rd] lm, weights and ...

[Peter Dalgaard]

"hadley wickham" <h.wickham at gmail.com> writes:

> > lm2 <- function(...) lm(...)
> > lm2(mpg ~ wt, data=mtcars)
> 
> Call:
> lm(formula = ..1, data = ..2)
> 
> Coefficients:
> (Intercept)           wt
>      37.285       -5.344
> 
> > lm2(mpg ~ wt, weights=cyl, data=mtcars)
> Error in eval(expr, envir, enclos) : ..2 used in an incorrect context,
> no ... to look in
> 
> 
> Can anyone explain why this is happening?  (Obviously this is a
> manufactured example, but it crops up in ggplot when you want to
> provide a function to perform smoothing)

Presumably this is due to the nonstandard evaluation of the weights
argument. This is a long story, the short version is that the argument
ought to have been designed to be a model formula, but we haven't been
willing to sacrifice S-PLUS compatibility. A more extensive
white-paper on the issues, written by Thomas Lumley, is at

http://developer.r-project.org/nonstandard-eval.pdf

-- 
   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907