bug#37352: 27.0.50; recursive-edit aborts on elisp error after evaluation

[Jean Louis <bugs@gnu.support>]

* Lars Ingebrigtsen <larsi@gnus.org> [2019-09-20 20:30]:
> Jean Louis <bugs@gnu.support> writes:
> 
> 
> [...]
> 
> >       (unwind-protect
> > 	  (recursive-edit)
> > 	(if (get-buffer-window buffy)
> > 	    (progn
> > 	      (setq new-value (buffer-substring (point-min) (point-max)))
> > 	      (kill-buffer buffy))))
> >       (switch-to-buffer this-buffer)
> >       new-value)))
> >
> > Then you could edit some string to see how this is happening:
> >
> > (read-from-buffer EDIT ME\nEvaluate this: (nonexistingfunction) )
> >
> > Then there after you could evaluate some non existing function to invoke
> > the Emacs Lisp error.
> >
> > At that time the editing with recursive-buffer is interrupted and data
> > is lost.
> 
> I'm not quite sure I understand the function above, are you basically
> doing this?
> 
> M-: (recursive-edit)
> M-: (error)
> 
> And you say that this returns you to the to the top level?
> 
> It doesn't do that for me -- unless I have debug-on-error set, and use
> `q' to exit that (which returns to top level, as advertised).
> 
> Is this what you're seeing?
> 
> -- 
> (domestic pets only, the antidote for overdose, milk.)
>    bloggy blog: http://lars.ingebrigtsen.no

Please see the attached picture. I have turned off debug on error
before evaluating (jnjn) non-existing function.

Jean