[R] period in scale parameter of prcomp

[William Dunlap]

> -----Original Message-----
> From: r-help-bounces at r-project.org 
> [mailto:r-help-bounces at r-project.org] On Behalf Of Greg Hirson
> Sent: Friday, May 13, 2011 10:58 AM
> To: R-help at r-project.org
> Subject: [R] period in scale parameter of prcomp
> 
> Hi all,
> 
> Looking at the help file for prcomp and the code in
> stats:::prcomp.default, the scale parameter for the default S3 method
> has a trailing period: it written as
> 
> scale.
> 
> I assume there is a reason for it (backwards compatibility with S was
> suggested) but I thought I'd ask.

In ancient versions of S you would get an error running
a function like
   f <- function(x, plot=TRUE) { if (plot) plot(x, x^2) }
because when it evaluated plot(x,x^2) the object called
'plot' was a logical vector, not a function.  The usual
workaround was to call the argument 'plot.', as in
   f <- function(x, plot.=TRUE) { if (plot.) plot(x, x^2) }
You could still call it as
   f(x, plot=FALSE)
since tag abbreviations are allowed in function calls.

Now the 'plot' in the function call 'plot(x,x^2)' is looked
up differently than the 'plot' in 'if(plot)': the former
search keeps going until it finds a function object.  Hence
the dot is no longer needed, although it might make you think
about how 'plot' can mean two different things in the same
environment.  Older functions still have this dot because
there is no good reason to change them and changing them can
break old code.

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com  

> 
> Thanks,
> Greg
> 
> -- 
> Greg Hirson
> ghirson at ucdavis.edu
> 
> Graduate Student
> Agricultural and Environmental Chemistry
> 
> 1106 Robert Mondavi Institute North
> One Shields Avenue
> Davis, CA 95616
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide 
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>