R help - Sep 2010 - advise on operations speed with Rcpp,Boost::ipc Shared Memory

Hi,

I have an implementation where I transfer data records via shared memory to an R
program. If anyone has prior experience, I'd like to find out which would be
faster

1) storing data records in shared memory as they are(in a matrix) and then use
the Rcpp::wrap to convert them to R datatypes
2) merge the records into a string and store records as strings. Then use R
functions like strsplit,lapply etc to convert them to their original matrix
form.

Any help is appreciated
	[[alternative HTML version deleted]]