integer functions {was [Rd] Inconsistencies (PR#550)}

maechler@stat.math.ethz.ch maechler@stat.math.ethz.ch
Mon, 22 May 2000 12:22:12 +0200 (MET DST)


>>>>> "BDR" == Prof Brian D Ripley <ripley@stats.ox.ac.uk> writes:

    BDR> On Mon, 22 May 2000 berwin@maths.uwa.edu.au wrote:
    >> Dear all,
    >> 
    >> I was playing around with some C-code that I dynamically linked to R
    >> and noticed the following inconsistency:

    > > dat <- matrix(1,3,3)
    > > is.double(dat)
    > [1] TRUE
    > > is.integer(dat)
    > [1] FALSE
    > > dat <- matrix(1:9,3,3)
    > > is.double(dat)
    > [1] FALSE
    > > is.integer(dat)
    > [1] TRUE

    >> How did I find out?  I passed `dat' to a double * in the C-code and
    >> printed out the matrix from C.  But in the .C() call I just passed
    >> down `dat' and forgot to coerce to double via `as.double(dat)'.  Of
    >> course the print out from the C routine was garbage, but since I used
    >>         dat <- matrix(1,3,3)
    >> first, I was confused.  Note that R says
    >> that `dat' is.double!  But it seems to pass dat to C as an integer
    >> matrix nevertheless!  Using the second matrix it became clear to me
    >> why I saw garbage being printed out from the C-routine.  My question
    >> is now why does the first `is.double(dat)' returns `TRUE'?

    BDR> Because it is double!  This refers to the storage mode, not the
    BDR> mathematical properties of the object's content.

exactly!  
Please (everyone who needs, probably Berwin doesn't) *do*
study  ?storage.mode  and  ?mode with some care

    BDR> I can find no inconsistency.  1 is double, and 1:9 is integer
    BDR> (just as in S-PLUS 3.x).  Why should this not be the case?

    BDR> Currently the only time in R you are likely to get an integer
    BDR> result is a : or seq expression, but you are not supposed to rely
    BDR> on this.

Well, "Values:"  in ?seq  says

 >> Value:
 >>      The result is of `mode' `"integer"' if `from' is (numerically
 >>      equal to an) integer.

on which I do rely..
Also, there are other functions guaranteeing an integer (or NA / NULL)
result, at least

 o length()
 o dim()
 o ncol(), nrow()
---
 o  min() and max()  --- when all args are integers

 o  "+" and "-" if the args are integers
    -- however that might change: Currently, there's no overflow detection!


    BDR>   (Since I know you have a copy, this is exactly the
    BDR> problem of the wam example on page 203 of `S Prograaming'.)

    BDR> For example, if printit.c is

        <...  instructive example...>

Martin Maechler <maechler@stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO D10	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._