Re: Accepting and returning multiple values in 'cl

[Dave Goel]

> The reason why I'm not sure we should install it, is that I'm afraid
>people got used to the current quirks of MRV, and the quirks of your
>alternative implementations aren't obviously preferable: yes, if we
>started cl.el now, I might choose your code over the current one, but
>I'm not sure it's worth the potential breakage now.

I do see your point.  I see a very few places in emacs, but it is
understandable.

Actually, I discovered cl's authors's own cl-compat.el, which is
already in emacs, which already seeks to do the right thing (just what
my cl-multiple.el did), with the alternative names like Values,
Multiple-value-*, etc.  I grepped through the whole emacs, nothing
uses these functions.


I see problems with cl-compat though, for example:


My code:
(multiple-value-list 
 (progn
   (values nil)
   (values)))
nil 


(Multiple-value-list 
 (progn
   (Values nil)
   (Values)))
(nil)
^^ that is wrong. 

Its implementation is also a bit funny and convoluted. It depends on
cl.el's multiple-value- implementations, which itself is hairy.

Do you mind if I fix these bugs with cl-compat.el and reimplement its
m-v functions?  There, I also want to add a note to the user: any time
they use forms that return m- values, the final form should invoke
(Values ) as well.

Keeping with cl-compat's usual terminology, cl-floor* and cl-truncate*
should be defaliased to Floor*, and Truncate*...

It seems like these changes won't affect the rest of emacs at all.