bug#30190: 27.0.50; term run in line mode shows user passwords

[Tino Calancha]

On Thu, 21 Jun 2018, Eli Zaretskii wrote:

> > From: Noam Postavsky <npostavs@gmail.com>
> > Cc: 30190@debbugs.gnu.org,  tino.calancha@gmail.com
> > Date: Wed, 20 Jun 2018 19:28:32 -0400
> >
> > Eli Zaretskii <eliz@gnu.org> writes:
> >
> > > I'm for fixing this in Emacs 26.2, but I still don't think I
> > > understand why the latest patch proposed in the discussion of this bug
> > > needs to "steal" so much from comint.el?
> > >
> > > Also, why does term-watch-for-password-prompt need to be invoked via a
> > > hook?
> >
> > I don't these things are really required; as far as I understand, Tino
> > did it that way in order to be safer: the "stealing" is to avoid loading
> > comint.el, and using the hook is to keep the code closer to the already
> > working example it's being copied from.
>
> Why is it a problem to load comint?  Either in this case or even
> always?
I have the bias/personal-preference to avoid load new things when I make a 
change.  Let's call it: 'disturb the least' with my patches.

> As for the hook: it looks strange to me to use hooks for this purpose,
> since IMO we are supposed to refrain from doing that as much as
> possible.
I must admit it: my patch brings cargo-cult from `comint.el'; comint.el 
performs such hook calls.

My patch just tried to mimic what is done in `comint.el' and reproduce
it in `term.el'.  My hope was that the patch would be accepted 
frictionless: if it's already done in `comint.el', 
why not doing the same in `term.el'?

The discussion turned out about point the implementation that we
have in `comint.el', which is also good and interesting topic.  A bit
out of scope of my initial intentions, but very welcome anyway.  Improve
code is always a good thing.

Stefan opinion on these 2 general questions might be very valuable.