[rfc/cli:rfa] Don't copy func() into show from set ..

[Andrew Cagney]

> Andrew, I agree with all that you say before but I have some
> additional thoughts.

I'll leave it for a week then commit it (if that is ok) .....

> Should we ever allow for the creation of a set command without
> a show?  If not, we should get rid of add_set_cmd() and
> add_show_from_set() and have a single add_setshow_cmd().
> This would prevent anyone of creating a set without the 
> corresponding show and hide the implementation details.

That's true.

> It could return the show pointer as an argument if so requested.

Ah, yes!  You're thinking of

	set = add_setshow_cmd (name, class, ..., &show);

where the &show could be NULL.

I can add it, and ARI (just info) the other versions, if you want.

Andrew