bug#17239: 24.3.50; competion error (cl-assertion-failed (string-prefix-p uprefix ustring))

[Stefan Monnier]

>> The resulting behavior is not great, but it's not really clear what we
>> should do anyway.
> As a point of comparison, when I do
> C-x C-f ~ / C-b TAB
> the behaviour of emacs depends on the current content of the minibuffer
> (which is default-directory by default I guess).

It shouldn't: it should complete the username (since the general form
of the ~ thingy is "~<user>/<filename>").

> I spotted these behaviours ("=>" means "hit C-b TAB and the minibuffer
> becomes") :
> ~/~/     => ~/~/ [Complete, but not unique]
> ~/foo/~/ => ~/
> /foo/~/  => /foo/~/ [Complete, but not unique]

Indeed, I'm not sure why ~/foo/~/ behaves differently.  It's probably
a side-effect of completion--sifn-requote.  You can take a look at that
function's leading comment to get an idea for why it might not always
behave perfectly ;-)


        Stefan