Possible Slowdown when i is an S4 Object with sapply method

[DCEmilberg]

Example (run on 1.9.6, but based on current master branch source the issue should be present there as well):

library(Matrix)
library(data.table)
x <- Matrix(1:1e6, nrow=1e3, ncol=1e3)
colnames(x) <- rownames(x) <- 1:1e3
dt.rowInd <- data.table(sink=1:900)
system.time(dt.rowInd[1:nrow(dt.rowInd),x[as.character(sink),]])
#   user  system elapsed 
#  0.027   0.004   0.031 
system.time(dt.rowInd[,x[as.character(sink),]])
#   user  system elapsed 
#  4.801   0.075   4.878 

This appears to be caused by code starting at line 1237 of datatable.R:

      if (is.null(irows)) {
            if (is.atomic(jval)) {
                jcpy = address(jval) %in% sapply(SDenv$.SD, address) # %chin% errors when RHS is list()
                if (jcpy) jval = copy(jval)
            } else if (address(jval) == address(SDenv$.SD)) {
                jval = copy(jval)
            } else if ( length(jcpy <- which(sapply(jval, address) %in% sapply(SDenv, address))) ) {
                for (jidx in jcpy) jval[[jidx]] = copy(jval[[jidx]])
            } else if (is.call(jsub) && jsub[[1L]] == "get" && is.list(jval)) {
                jval = copy(jval) # fix for #1212
            }
        }

In particular:

sapply(jval, address) %in% sapply(SDenv, address)

Because jval in this case is an S4 object logic is such that the line above contained in the third level of the if/elseif statement is run since is.atomic(jval) is not true. Additionally, since Matrix defines an as.list method, sapply(jval, address) runs address on every single element of the Matrix, which results in a massive slow down. A possible fix might be to change line 1238 to:

if(!is.list(jval)) {