Chris Gregory wrote:
> I'll try to benchmark it
> tomorrow. If it inlines the function to Fprogn then there is no benefit to
> performance (assuming it elides the Lisp_Object completely). The code,
> however, is cleaner now.

Thanks, I looked into this and there's no need to benchmark, as the change (when 
fixed up) does not alter the generated machine code, at least on my platform 
(Fedora 24 x86-64, GCC 6.3.1). The basic idea does simplify and clarify the code 
a bit, so I installed the attached.