bug#23949: 25.0.95; Regression in handling error caused by (string-match-p "." nil)

[Eli Zaretskii]

> From: Kaushal Modi <kaushal.modi@gmail.com>
> Date: Wed, 13 Jul 2016 13:10:46 +0000
> Cc: Eli Zaretskii <eliz@gnu.org>, schwab@suse.de, 23949@debbugs.gnu.org
> 
> So what is the way forward?

I just proposed one way forward.

> Certainly there was a lower level change after emacs 24.5 that changed the 
> behavior of how the run-hooks
> behave or how the advices are executed in general or when called within a 
> debugger? 

The underlying program with binding inhibit-changing-match-data when
help-function-arglist is called exists on the emacs-25 branch as
well.  It also exists in Emacs 24.5, except that there
help-split-fundoc signals an error, instead of returning bogus data:

  (let ((inhibit-changing-match-data t))
    (help-split-fundoc (documentation 'delete-file) nil))
    => error->Args out of range: 3004