[R] question from Braun/Murdoch book

[Duncan Murdoch]

On 08/10/2008 1:48 AM, Erin Hodgess wrote:
> Hi R People:
> 
> I am looking at the Braun/Murdoch book, " A First Course in
> Statistical Programming in R", and I have a question about a function
> there.  It's on page 52, Example 4.5; the sieve of Erastosthenes.
> 
> There is a line:
> primes <- c()
> 
> Is there a difference between using that and
> primes <- NULL
> please?
> 
> When you put in primes <- c(), primes comes back as NULL.
> 
> 
> Is one more efficient or is this just a matter of programming style, please?

It was purely a choice of style.

By the way, there is an error in one of the programs coming soon after 
that:  the mergesort example that starts on p. 68 doesn't handle 
odd-length input properly, because it uses "len / 2" in a number of 
places where it really needs "len %/% 2".  So for example, when len is 3 
we get behaviour like this:

 > x <- 1:3
 > len <- 3
 > x[1:(len/2)]
[1] 1
 > x[(len/2 + 1):len]
[1] 2

(and the 3 got lost).  Using the integer divide is fine:

 > x[1:(len %/% 2)]
[1] 1
 > x[(len %/% 2 + 1):len]
[1] 2 3

Duncan Murdoch