bug#58537: CC Mode 5.35.1 (C/*l); Keywords being fontified as types at random

Hello again, Po.

On Wed, Oct 19, 2022 at 18:27:34 +0800, Po Lu wrote:
> Alan Mackenzie <[email protected]> writes:

> > I think the patch solves #58537 together with the problems you reported
> > since.  Does it also solve #58539?

> Yes, it does, ....

That's fantastic!

>  .... but there's still an unsolved problem:

> test ()
> {
>   dm_window_ptr window;

>   window = id_private_fetch ();
>   if (window)
> }

> if this is edited into:

> test ()
> {
>   dm_window_ptr window;

>   window = id_private_fetch ();
>   if ((window)addr window)
> }

> which is not valid C, but still something that can happen while typing a
> program, and it is then edited into:

> test ()
> {
>   dm_window_ptr window;

>   window = id_private_fetch ();
>   if ((window_addr) window)
> }

> "window" in the line above the "if" remains fontified as a type.

Sorry, you mentioned this before, and I didn't get around to fixing it.

Would you please try applying the attached patch on top of the earlier
patch from this morning (for me)/afternoon (for you).

If this works OK, then I propose linking bugs #58537 and #58539 and
closing them.  Thanks for being so patient and insistent over the last
few days.

> Thanks.

-- 
Alan Mackenzie (Nuremberg, Germany).

diff -r dbe2661a4298 cc-engine.el
--- a/cc-engine.el	Mon Oct 17 20:12:54 2022 +0000
+++ b/cc-engine.el	Wed Oct 19 12:33:58 2022 +0000
@@ -11143,7 +11161,10 @@
 		    (not (c-on-identifier)))))))))
 
       ;; Handle the cast.
-      (when (and c-record-type-identifiers at-type (not (eq at-type t)))
+      (when (and c-record-type-identifiers
+		 at-type
+		 (not (memq at-type '(t maybe)))) ; 'maybe isn't strong enough
+					; evidence to promote the type.
 	(let ((c-promote-possible-types t))
 	  (goto-char type-start)
 	  (c-forward-type)))