set-variable and customize-set-variable

[Drew Adams]

1. It's not clear to me why we have two commands for setting a user option,
`set-variable' and `customize-set-variable', and why their user interfaces
are so different.

The doc strings for these two commands are very similar. IIUC, the main
difference is that s-v sets the current value only, and c-s-v sets the
default value. I might not understand the difference 100% well - what is it?
When do we expect users to use one or the other command?

Can/should these two commands be combined somehow? Each already lets you use
a prefix arg to alter its behavior slightly (local value for s-v, add a
comment for c-s-v). Why not merge the two commands somehow?


2. As long as we do have both commands:

 - I don't find any mention of c-s-v in the Emacs manual (or the
   Elisp manual).

 - Can we change the UI of s-v so that it too allows completion
   when appropriate (e.g. for a `choice' that includes `const's)?
   If completion can be used for c-s-v, then why not also for s-v?

Of course, completion would only be available for user options defined with
defcustom, not with defvar (with `*' starting the doc string). That is,
completion would only be available when the possible values were known. IOW,
why not let s-v do what c-s-v does when it can do so?

The convenience of completion makes it all the more questionable that c-s-v
is not even mentioned in the manual.