bug#50236: 27.2; electric-pair-mode is inconvenient in comint

On Mon, Aug 17, 2026, 11:56 Eli Zaretskii <[email protected]> wrote:

> .
> >
> > Yes, indeed in the interest of getting this bug fix moving and hopefully
> > into emacs-31, lets:
> >
> > 1. Push the attached patch to emacs-31 (this is a bug fix after all).
> > Give it some testing.  I'll be testing it myself.
> > If there are no objections, I'll push in a few days time.
>
> Is this bug new in Emacs 31?


Definitely not new. Quite old in fact. But very real. I've worked around it
in my SLY extension which  used to be the main use I did of comint. Other
than that, I live with the pain and/or hope the process output doesn't
output unbalanced things.

Comint.el is still very heavily used today, for eshell, REPLs, and anything
prompt-based (like those agentic LLM interfaces).

In parallel, my perception is that after misunderstanding it for a while
after Emacs 24 came out with it, people have slowly learned to appreciate
e-p-m's balancing, and leaving smartparens.el behind in favour of it. It
makes sense that they want e-p-m it to "just work" in comint as it largely
does everywhere else.

I think this fix would make a fine addition to Emacs 31's polish, and, as I
explained in my previous email, the mechanism has plenty of room to grow
into more use cases (in master, naturally).

If not, I'm not sure it is wise to
> install on the emacs-31 branchrvof so close to the release.  Sean, WDYT?
>

Should be reasonably safe, is my judgement. It's very elegant and small.
Also I made it a noop in anything but comint.el modes. Comes with tests.
But your call!

>
João

>