[R] %+=% and eval.parent()

[Peter Dalgaard]

Robin Hankin <rksh at soc.soton.ac.uk> writes:

> Some time ago, Peter Dalgaard made the wonderful suggestion to define
> 
> "%+=%" <- function(a,b) {eval.parent(substitute(a <- a + b)) }
...
> R> a <- matrix(1:9,3,3)
> R> a[a%%2==1] %+=% 1000*(1:5)
> [1] 1001 2006 3015 4028 5045
> R> a
>       [,1] [,2] [,3]
> [1,] 1001    4 1007
> [2,]    2 1005    8
> [3,] 1003    6 1009
> R>
> 
> Why the difference when I remove the brackets?
> And how is the vector that is printed in the unbracketed version determined?
> (why is this printed anyway? doing "a[a%%2==1] %+=% 1000" doesnt
> print anything).

You're getting caught out by operator precedence:

a[a%%2==1] %+=% 1000*(1:5)

is effectively

(a[a%%2==1] %+=% 1000)*(1:5)

because %foo% operators have higher precedence than * (and much higher
than ordinary assignment)

> Is there a better way to code up %+=%  ?

Not really. You could hack the parser, but from within R, there is no
way to set parser precedence.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907