R devel - Apr 2014 - vapply confusion

The following code seems to contain an inconsistency in
the behavior of vapply(). Am I missing something here?

## This function assumes v is a 3d vector, beta a scalar.
f3d <- function(v,beta) { v+beta }

## This expression applies f3d to a vector of scalars, and
## specifies the template 'array(10,3)' for the return value.
dat <- vapply(seq(0,1,length=10), function(beta) {
  f3d(c(0,0,0), beta) }, array(10,3))

dim(dat) # 3 10 (the dimensions are the reverse of the template?)

Of course, a transpose will patch the result to match the
specified template.

Dominick Samperi <djsamperi <at> gmail.com> writes:
> 
> The following code seems to contain an inconsistency in
> the behavior of vapply(). Am I missing something here?
> 
> ## This function assumes v is a 3d vector, beta a scalar.
> f3d <- function(v,beta) { v+beta }
> 
> ## This expression applies f3d to a vector of scalars, and
> ## specifies the template 'array(10,3)' for the return value.
> dat <- vapply(seq(0,1,length=10), function(beta) {
>   f3d(c(0,0,0), beta) }, array(10,3))
> 
Same result as that of sapply(...
> dim(dat) # 3 10 (the dimensions are the reverse of the template?)
> 
No the template has dimension '3'
> dim(array(10,3))
[1] 3
> 
and X has length 10.

Editting the help page:

"If FUN.VALUE is [an array, the result is] an array a with dim(a)
=c(dim(FUN.VALUE), length(X))."

which is what you got.

HTH,

Chuck