[Rd] proposed changes to RSiteSearch

[Wacek Kusnierczyk]

Romain Francois wrote:
> strapply in package gsubfn brings elegance here:
>
> > txt <- '<foo>bar</foo>'
> > rx <- "<(.*?)>(.*?)</(.*?)>"
> > strapply( txt, rx, c , perl = T )
> [[1]]
> [1] "foo" "bar" "foo"
>

sure, but this does not, in any way, make it less strange that gsub is
not vectorized. 


> Too bad you have to pay this on performance:
>
> > txt <- rep( '<foo>bar</foo>', 1000 )
> > rx <- "<(.*?)>(.*?)</(.*?)>"
> > system.time( out <- strapply( txt, rx, c , perl = T ) )
>   user  system elapsed
>  2.923   0.005   3.063
> > system.time( out2 <- sapply( paste('\\', 1:3, sep=''), function(x){
> + gsub(rx, x, txt, perl=TRUE)
> + } ) )
>   user  system elapsed
>  0.011   0.000   0.011

    strapply

and you know why.


vQ