🚀 go-pugleaf

GNU-Emacs 18.55 :

I think that there's a bug in scan_sexps_forward () in syntax.c.

The bug manifests it self if you have a line that has a close paren as the
first non-white space, point is before the close paren, and you execute

    (parse-partial-sexp (point) (point-max) -1)

The return value has a bogus element at position 3. The problem appears to
be that if the first case label selected in the main switch is case Sclose,
the statement

    curlevel->prev = curlevel->last; /* ~ line 947 */

is bogus, since curlevel->last is an auto var, and has not been set. Even
though curlevel->prev is set to -1 at the start of the function,
curlevel->last is not. I suspect that setting it along with curlevel->prev
would solve the problem. I've put that fix in Epoch, and it seems to work.