Hi,

While I'm working at the REPL, I sometimes need to use symbols that are not present in the current package. In this case, the slime-complete-symbol* function cannot complete the symbol at point.

Here I propose a small change in slime.el in order to ask the user whether he/she wants to try completions in different packages.

< (if (null completion-set) < (progn (slime-minibuffer-respecting-message ---

(if (and (null completion-set)
         (if (y-or-n-p (format "Can't find completion for \"%s\" in the current package. Search other packages?" prefix))
           (dolist (package (slime-eval `(swank:list-all-package-names t)) t)
             (setq completion-result (slime-completions (format "%s::%s" package prefix))
                   completion-set (first completion-result)
                   completed-prefix (second completion-result))
             (when completion-set
               (return nil)))
           t))
    (progn (slime-minibuffer-respecting-message 

I used it, mainly, to "automagically" insert the package qualifier on a given function call.

It works but it could be improved. The problem is that it stops searching packages as soon as one of them returns some completions. I tried to adapt it to present the user with all possible completions in all packages but the format of the slime-completions returned values and the use of the completed-prefix variable (that assumes that there's only one package) makes this a bit harder.

So, before I start making some more serious changes, I would like to know if anyone thinks this is sufficiently useful to deserve more time investment.

Best regards,

António Leitão.

Ps: on the Explorer Lisp Machine, this was _very_ useful when we were debugging large systems. But the behaviour wasn't exactly like this: IIRC, it was implemented using restarts.