error/crash with subset on empty tables where zeros are at the end of the non-empty subset

[arunsrinivasan]

I've managed to reproduce this on Windows. Here's a MRE.

require(data.table) # v1.9.8, R version 3.3.1, Windows 7
.Call("CsubsetVector", character(0), 1:0)
# Error: attempt to set index 1/1 in SET_STRING_ELT

Running it again hangs up the R session. The actual usage was subsetting of an empty table, but I tracked it down to subsetVector. Note that both .Call("CsubsetVector", character(0), 1L) and .Call("CsubsetVector", character(0), 0L), from a fresh session, work fine.