Reproduction: [1 2] dup first .s
Expected output: [ [ 1 2 ] 2 ]
Actual output: [ [ 2 -0.00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000037206620809969885 ] 1 ]

I'm not sure whats causing the weird float in the output, but whats happening is obviously some kind of "copy by reference" bug, especially because [1 2] dup rev rev first .sworks as expected.

While on the subject, may I suggest adding a pop_from_other_side command (aka the opposite of enq), as that is what I was trying to do here but it probably should be a builtin.