[Bug c/89990] request warning: Use of out of scope compound literals

["modchipv12 at gmail dot com" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89990

--- Comment #7 from Andrew D'Addesio <modchipv12 at gmail dot com> ---
(In reply to Andrew Pinski from comment #6)
> The warning is now included in GCC 12.
> And this makes this a dup of bug 63272.
> 
> *** This bug has been marked as a duplicate of bug 63272 ***

Yep, the new warning is working nicely on my test case (comment #5) on GCC
14.0.1 on Fedora 40 x86-64:

    test.c: In function ‘test’:
    test.c:22:15: warning: dangling pointer ‘ptr’ to an unnamed temporary may
be used [-Wdangling-pointer=]
       22 |     return ptr->c;
          |            ~~~^~~
    test.c:15:37: note: unnamed temporary defined here
       15 |         ptr = &(const struct mytype){ 43 };
          |                                     ^

Though one thing to note -- (in addition to the above warning) it still
generates that "strange" -Wuninitialized warning if and only if there are one
or fewer calls to foo():

    In function ‘test’,
        inlined from ‘main’ at test.c:27:12:
    test.c:22:15: warning: ‘<Ufe10>.c’ is used uninitialized [-Wuninitialized]
       22 |     return ptr->c;
          |            ~~~^~~

^ If foo() is called 2+ times then that warning disappears.

It's not a bug per se as the "real" -Wdangling-pointer warning still gets
displayed to the user (Ufe10 is probably GCC's representation of the variable
that was never initialized).

But I find it very peculiar that the threshold for silencing that
-Wuninitialized warning is 2 function calls and not say 1 or 10.

This actually has gotten me curious. Would you have an idea/explanation behind
that 2 function call threshold @Andrew Pinski?