I claimed to have written over a million lines of code, perhaps my claim was bogus, and I should have said that I caused 1 million lines of code to be written? I bet that many of the people who claim have written large amounts of code write code generators of some sort. I have written several that have targeted fortran, assember, C, CLR and JVM. If you want great KLOCS, generating fortran and assembler from a MIL is pretty hard to beat. :-)
My personal record was over 500K lines of assember in a day to add/remove instrumentation to a previously built system. The data we wanted from the function calls was quite specific, so we couldn't easily get what we needed from valgrind or quantify. Since the code was compiled and linked into a product, and I was the sole author of the analyser and generator, I consider that a 500 KLOC day. :-)
Bogus, perhaps, but the source control system had my name on it. :-)
Counting the code written by my generators may be a tad optimistic, but that is probably what most of the million LOC people who weren't totally lying did. But I suppose the pessimistic pedants of you would probably say folks like us should only count the code in the code generator. Instead of counting what got checked into source control....