Yes, the answer I would like to hear is that in a high level language - if the language itself doesn't support it (e.g. in Ruby you can do 'a, b := b, a' or a p

Yes, the answer I would like to hear is that in a high level language - if the language itself doesn't support it (e.g. in Ruby you can do 'a, b := b, a' or a parallel computing language) - you don't fight the compiler but use the correct idiom that the compiler can optimize. You can still say that if you are programming in assembler or java bytecode for optimization reasons you can do any of those stupid programming tricks.