__STDC__ == 0 in System Headers on Solaris?

["Jeffery Cavallaro" <jeffery at cavcom dot com>]

I didn't get any response on gcc-help.  Hopefully the developer
community can let me know what is going on!

-------------------------------------

Hi, all - need a little help with gcc 3.2.2, __STDC__, and the solaris
2.9 system header files.

If __STDC__ != 0, then <sys/types.h> will define longlong to be a
composite type, which messes up all of the large file stuff.  In past
GCC versions, I have been successful is redefining __STDC__ to 0 so that
everything works correctly.  OK, I know, very kludgy, but I didn't see
any other choice at the time.  Now, GCC is VERY stringent about __STDC__
redefinition, and I don't see an out without putting up with reams of
__STDC__ redefined warning messages (nor do I wish to patch gcc to get
rid of the message).

When I dealt with this is past versions, I noticed a note in the GCC
documentation stating that on systems with a convention of __STDC__=0,
GCC would use a 0 value when scanning system header files.  Indeed, this
feature seems to be enabled in the Solaris 2.x support
(STDC_0_IN_SYSTEM_HEADERS, or some such define).  Yet, this doesn't seem
to be having any effect - __STDC__ is still 1.

Am I completely misunderstanding this feature?  Help please!  I would
like to do this correctly and have longlong be defined as the correct
atomic type (long long).

 

Thanks!

Jeff
Please reply directly (I am not on the developers' list)