bug#51982: Erroneous handling of local variables in byte-compiled nested lambdas

["Mattias Engdegård" <mattiase@acm.org>]

> Yes, and I can confirm that both transformations (let* -> let and the inverse) seem to generate almost or exactly identical bytecode for lexbind code. If you can find an example where it's worse, I'd like to know.

And by "yes" I mean "no", since the identical bytecode is only true in the absence of mutation which, as always, throws a spanner in the works. For example, while

(let ((x (sin y)) (y (cos x)))
  (+ x y))))
->
(let* ((x0 (sin y)) (y (cos x)) (x x0))
  (+ x y))))

is a transformation with identical bytecode, mutating one of the variables:

(let ((x (sin y)) (y (cos x)))
  (setq x (1+ x))
  (+ x y))))
->
(let* ((x0 (sin y)) (y (cos x)) (x x0))
  (setq x (1+ x))
  (+ x y))))

will prevent the intermediate from being eliminated. Not a disaster by any means but not good enough as a general method of translation.