> Why add length= as a primitive? Why is it important enough to justify that?

About 14% of all calls to length are then simply compared via =. I checked and the < and > variants combined account for about 30%, the <= and >= combined about 10%. I don't really see a purpose in using nthcdr like you have in a lisp implementation of length=, it isn't any better than just using length directly since it requires just as much linked list traversal. That was of course the reason I implemented it in C.

> Wouldn't it make sense to have a function ‘length<’ and ‘length>’ as well?

Yes, it would appear so. But for the same reasons as length= they would need to be implemented in C for there to be any improvement in efficiency. They could be added to the < and > bytecodes in the same way that I've done with = and length=.