Re: autoconf-2.52g on HP-UX (was: bison 1.32)

[Paul Eggert]

> >   # The cast to unsigned long works around a bug in the HP C Compiler
> >   # version HP92453-01 B.11.11.23709.GP, which incorrectly rejects
> >   # declarations like `int a3[[(sizeof (unsigned char)) >= 0]];'.
> >   # This bug is HP SR number 8606223364.
> 
> Do you know if the report is available on SUM, or only to the reporting
> customer? If public, through what interface?

Sorry, I don't know.  That SR number was reported by another HP user,
who said that a compiler guy at HP had filed a bug report.

> Could you please be more specific.

That's not necessary, as you answered the question indirectly in your
later answers.

> l1:/tmp 115 > cat conftest2.c
> int main ()
> {
>     static unsigned char uc;
>     static int test_array[1 - 2 * !((sizeof uc) >= 0)];
>     test_array[0] = 0;
>     return (0);
>     } /* main */
> l1:/tmp 116 > cc -c -Ae +O2 +Onolimit +DAportable -I/pro/local/include 
> -I/usr/include/X11R6 conftest2.c
> cc: "conftest2.c", line 4: error 1879: Variable-length arrays cannot have 
> static storage.

Yes, it's clear that that cc is buggy.

Is there some way to work around the problem?  What happens if you
compile with plain 'cc -c', for example?

Also, how about the following three programs?  I'm asking because the
HP bug report indicates that at least the first program should work.

 static int test_array[1 - 2 * !((int)(sizeof(unsigned char)) >= 0)];
 int main ()
 {
    test_array[0] = 0;
    return (0);
 }

 static int test_array[1 - 2 * !((long)(sizeof(unsigned char)) >= 0)];
 int main ()
 {
    test_array[0] = 0;
    return (0);
 }

 static int test_array[1 - 2 * !((unsigned long)(sizeof(unsigned char)) >= 0)];
 int main ()
 {
    test_array[0] = 0;
    return (0);
 }