Re: [bug-gnulib] Re: lib/stdbool_.h doesn't honor HAVE__BOOL

[Paul Eggert]

Bruno Haible <address@hidden> writes:

>>  /* BeOS <sys/socket.h> already #defines false 0, true 1.  We use the same
>> -   definitions below, but temporarily we have to #undef them.  */
>> +   definitions below, which is OK.  */
>>  #ifdef __BEOS__
>>  # include <OS.h> /* defines bool but not _Bool */
>> -# undef false
>> -# undef true
>>  #endif
>
> Do you have a machine on which you tested this part? I mean, I wouldn't
> have put in the #undefs if I hadn't seen gcc warnings otherwise.

I didn't think they were necessary, given the other changes I proposed
that removed the enums defining true and false.  If we keep those
enums, though, then the #undefs have to stay too.

>> +      #if defined __xlc__ || __GNUC__
>> +       /* Catch a bug in IBM AIX xlc compiler version 6.0.0.0
>> +          reported by James Lemley on 2005-10-05; see
>
> If this is meant to catch a bug in XLC, what's the purpose of the
> __GNUC__ conditional?

To test that nobody screws up the check in the future.  GCC passes
that test, and should always pass that test; so if GCC fails then
there's something wrong with that test.

I'll look at your other comments and send in a proposed revised patch
which will at least put in a comment about why that __GCC__ is useful.
There are still some problems at least on HP-UX compilers, I think,
and I'll try to address those too.