bug#61162: 30.0.50; RET in source files that are being displayed by gud no longer works

[Eli Zaretskii <eliz@gnu.org>]

> From: Michael Welsh Duggan <mwd@md5i.com>
> Date: Tue, 31 Jan 2023 14:17:20 -0500
> 
> Michael Welsh Duggan <mwd@md5i.com> writes:
> 
> > Create a trivial C program.  Here's a good example:
> >
> >     int main()
> >     {
> >       return 0;
> >     }
> >
> > Save it as foo.c, and compile it with debugging:
> >
> >     gcc -ggdb3 -o foo foo.c
> >
> > From emacs -Q, start gdb on foo:
> >
> >     M-x gud-gdb RET 
> >
> > In my case, since foo is the only executable in the dir, the command
> > line for gdb is already set up correctly as:
> >
> >     gdb --fullname foo
> >
> > Hit RET.  You should be at the gdb prompt.  Start the program:
> >
> >     start RET
> >
> > You should now get another window with foo.c in it for context.  Switch
> > to that buffer and type RET.
> >
> >     C-x o
> >     RET
> >
> > In my case, I get:
> >
> >     comint-send-input: Current buffer has no process
> >
> > This is new and broken behavior.  The expected behavior is that a
> > newline is inserted in the buffer.
> 
> This appears to be caused by commit
> a7b0b20c1bc758fa99fda96d06a266868aec3eea, which was supposed to fix
> Bug#59605.

This current problem is part of fallout from commit 8bb5c1bfec.  That
commit was reverted on the emacs-29 branch, as it caused too much
trouble.  It was not reverted on master, as I still hope someone
motivated enough will step up to fix this without reverting back to
using easy-mmode, which we hoped to obsolete.

If you can find a fix on master, patches would be very welcome.
Failing that, I might eventually decide to revert that commit on
master as well.  There's a limit to the efforts I'm prepared to sink
into maintaining what is basically an obsolete package which uses a
GDB feature that GDB developers consider obsolete and may remove any
day (why don't you use "M-x gdb" instead?).