bug#39389: 27.0.60; A couple of bugs messing with minibuffer completion of /sudo::

[Jimmy Yuen Ho Wong <wyuenho@gmail.com>]

On Sun, Feb 2, 2020 at 8:11 PM Michael Albinus <michael.albinus@gmx.de> wrote:
>
> Jimmy Yuen Ho Wong <wyuenho@gmail.com> writes:
>
> >> I'm just surprised, that you set ido-enable-tramp-completion to nil, and
> >> that you still expect Tramp file name support. That doesn't sound right
> >> to me.
> > I expect either that variable didn't exist and Tramp file name
> > completion was always fast, or when it's off, ido just reverts back to
> > the behavior of the regular `find-file`, which should allow me to type
> > out the whole Tramp syntax.
>
> Why do you expect this? At least the docstring of that variable doesn't
> tell that ido behaves like this, when the variable is set to nil.
>

Because no reasonable person will knowing turn on something that will
break tramp during find-file completely.

> I agree with you, that the variable shall document its behaviour in this
> case.
>

It's not a documentation problem. It's called the principle of least
astonishment, and the culprit for bug 1 is hidden somewhere in
`ido-read-internal` in one of the 2 cases that deals with tramp
syntax.

For bug 2 and 3, there's definitely something wrong with the
`auth-source-search` call in `tramp-read-passwd` that's blocking the
display of the password prompt, as the same workaround works for both
issues.