On 9/26/2021 1:58 PM, Alan Mackenzie wrote:
> 
> There's one thing I'm a bit uncertain about in the patch for cc-mode.el.
> 
> On Wed, Sep 22, 2021 at 19:01:11 -0700, Jim Porter wrote:
[snip]
>> -  (if (eq char ?\")
>> -      (not (equal (get-text-property (1- (point)) 'c-fl-syn-tab) '(15)))
>> -    (funcall (default-value 'electric-pair-inhibit-predicate) char)))
>> +  (or (and (eq char ?\")
>> +	   (not (memq (cadr (c-semi-pp-to-literal (1- (point)))) '(c c++)))
>> +	   (let ((last-quote (save-match-data
>> +			       (save-excursion
>> +				 (goto-char (c-point 'eoll))
>> +				 (search-backward "\"")))))
>> +	     (not (equal (c-get-char-property last-quote 'c-fl-syn-tab)
>> +			 '(15)))))
>> +      (funcall (default-value 'electric-pair-inhibit-predicate) char)))
>   
> In the line starting (or (and (eq char ?\"), don't we still need an `if'
> form?  I think that otherwise, if any of the sub-forms of the `and'
> return nil, we will call (default-value
> 'electric-pair-inhibit-predicate), which surely isn't what we want.  If
> we have a ", we want the CC Mode function to do all the work, only
> delegating to the standard function when we don't have a ".
> 
> Or have I missed something?

I don't have a strong opinion on this either way, so here's a patch that 
uses an `if' form as you suggest. However, I introduced the `or' form 
in response to Eli's concern[1]:

On 9/11/2021 11:26 PM, Eli Zaretskii wrote:
> Your expected results seem to expect Emacs to assume that a new
> string will be inserted, but is that an assumption that is always
> true?  It could be that the user wants to modify the existing string
> instead, in which case your suggested patches will require the user
> to delete more quotes than previously.

I discussed the pros and cons in my followup[2]:

> Note however that this solution isn't perfect: it means a user's
> custom `electric-pair-inhibit-predicate' can only inhibit *more*
> than CC mode's default behavior, not less. I think that's a
> reasonable compromise though, and users who want more direct control
> can [override `c-electric-pair-inhibit-predicate'].[3]

Personally, I'm fine with letting CC Mode do all the work here, and 
requiring users to advise this function if they want a different 
behavior. It's probably easier to be sure that things don't break by 
using the `if' form patch, though I think the previous `or' form patch 
should be pretty robust too.

[1] https://lists.gnu.org/archive/html/bug-gnu-emacs/2021-09/msg00976.html
[2] https://lists.gnu.org/archive/html/bug-gnu-emacs/2021-09/msg01014.html
[3] My explanation in the original message was incorrect, so I've fixed 
it here.