bug#38992: 27.0.60; when enabled, fido-mode seems to break vc-git-grep

[Dmitry Gutov]

On 05.03.2020 0:44, João Távora wrote:
> fido-mode doesn't change icomplete-mode. It shouldn't, at least.
> It just uses it as a library.  And surely if it's a bug in fido-mode,
> surely it needs fixing _there_  and not elsewhere.  But it seems
> not be the case (at least with your latest proposal), so you have
> me confused.

Since the problem isn't triggered by icomplete-mode, but is triggered by 
fido-mode, it seems the latter binds some commands that are not a great 
fit for it.

Hence the second patch which added a new command specifically for fido-mode.

The first one changes icomplete-force-complete-and-exit to honor 
require-match=nil, failure to do which wasn't a problem for 
icomplete-mode because one uses C-j in incomplete-mode only to pick 
among one of the suggested matches. ido-mode users, however, like to use 
RET for arbitrary inputs as well.

If I were to classify, the first one fixes a "UI deficiency", but the 
second one fixes a bug.

> So much time has passed that I don't remember what the fix was anymore.

The previous patches are attached to the older message. Not too hard to 
find on the debbugs bug page.

> Have you pushed it?  I don't see it in emacs-27

Just got back to this discussion.

I've pushed the updated (more limited) patches to the 'fido-mode-fix' 
branch.