the additions to implement nested named graphs have grown to about 260 lines of code. much of this has to do with interpreting the variant inclusion classes together with handling the dataset clause
berlinnew:spocq-seg james$ wc repository-streaming.diff 180 1078 10867 repository-streaming.diff berlinnew:spocq-seg james$ wc classes.diff 64 271 2838 classes.diff berlinnew:spocq-seg james$ wc bgp-versioned.diff 18 43 1066 bgp-versioned.diff
repository-streaming is the core of the match/scan mechanism upon which the sparql processor is built. notable about its changes is that they are solely to compute the set of graph identifiers which define the effective target graph.