Re: Does `side-effect-free` have to apply to sub-functions too?

> I am wondering whether a function being free of side effects applies to
> functions that use `seq-elt`, which could run an unknown method 
> implementation for an unknown custom type. If it is possible for the 
> unknown method implementation to have side effects, then is the function 
> which unknowingly triggers the unknown implementation considered to not 
> be side-effect free?

Marking it as side-effect-free should be taken as a declaration
of intent.  I.e. if it turns out that in some cases it does have
a side effect, then we have a bug.  At that point we can look at the
details to determine if the bug is in the declaration or in
the implementation.

> In other words, before marking a function as `side-effect-free`, must 
> one check the function's sub-functions, and those sub-functions' 
> sub-sub-functions, and so on, all the way down, even if the function 
> itself does not intentionally have a side effect?

It's simply impossible to "check the function's sub-functions ..."
(consider `advice-add` if you're not convinced).


=== Stefan