[R] options(): "no no" {was `Subsetting a matrix'}

[Martin Maechler]

>>>>> "Ted" == Ted Harding <Ted.Harding at nessie.mcc.ac.uk>
>>>>>     on Mon, 14 Jul 2003 13:50:35 +0100 (BST) writes:

    Ted> On 14-Jul-03 Adelchi Azzalini wrote:
    >>> Maybe it is reasonable to propose incorporating "drop"
    >>> as one of the things you can set with "options"? :--
    >>> [...]
    >>  Perhaps this is a way-out. As usual, everything is
    >> feasible when we resort entirely on our own code.  The
    >> danger is that we might make use of some package (may be
    >> written some years ago) which assumes drop=TRUE is the
    >> default behaviour, and changing the default behaviour
    >> might (potentially) have disasterous effects.

    Ted> Indeed. However, it should be OK when embarking on a
    Ted> block or function of matrix computations, written by
    Ted> oneself, to enclose it between

    Ted>   options(drop=FALSE) ...  options(drop=TRUE)

    Ted> (unless some of the standard or "base" matrix functions
    Ted> which one might use in "..." depend on the default. How
    Ted> could one tell?).

First:  
   Yes, most if not all good S programmers I know agree that
   drop = FALSE should have been the default --- when S was created.
So you are right, here, Adelchi.

But I'm pretty confident that we, the R core team, pretty
unanimously believe that  options() that influence the result of
computations are  `X' ,  where X is in
  {"bad", "evil", "design flaw", "deserves death penalty"}   (;-)

Currently, the only official exception is the contrasts setting;
all other options don't influence computations, but ``output''
(printed and graphical) and interface to R-external tools
including `input' such as character encoding and `ts.eps' for
time-series rounding behavior.

Martin Maechler <maechler at stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><