On Nov 9, 2003, at 4:20 PM, Albert Chin wrote:
On Sun, Nov 09, 2003 at 02:19:40PM -0800, Guy Harris wrote:
On Sun, Nov 09, 2003 at 04:07:35AM -0600, Albert Chin wrote:
1. HP C compiler complains about typedef before enum declared
(HP-UX 10.20).
Checked in. (Is that legal ANSI C, or is that just an HP C compiler
deficiency? My ANSI C standard is at work....)
Dunno. My copy is at work too :)
I suspect it might not be legal ANSI C, so the HP compiler might be
allowed to reject it. I don't think ANSI C allows
typedef struct foo foo_t;
without at least an incomplete definition of "struct foo", and the same
would presumably apply to enums.
3. P_SID in <sys/procset.h> on HP-UX 10.20 and Tru64 UNIX.
I checked in a change a few days ago to get rid of includes of
<gmodule.h>, <sys/types.h>, and <netinet/in.h>; I don't think they're
needed, and they might keep <sys/procset.h> from being included.
See whether getting rid of those #includes, and reverting back to
P_SID,
works.
I copied the latest packet-ansi_a.c from CVS and still get the error.
I'm curious what causes all that stuff to be included, in case there's
some #include we can get rid of. Any idea?
4. HP C compiler on HP-UX 10.20 complains about the initializer for
ett ("Array size must be a constant expression.").
At least as I read that code, it *is* a constant expression; am I
missing something, or is this a deficiency of that C compiler?
Looks like a deficiency of the compiler. This works:
static gint *ett[sizeof(ansi_dtap_strings)/sizeof(value_string)]
But this does not:
static gint *ett[sizeof(ansi_dtap_strings)/sizeof(value_string) < 0
? 0 : 1]
Sigh. I've checked in your patch to work around that deficiency;
unless I'm missing something, the ANSI C89 spec says the array size in
the latter example *is* a constant expression. If the latest version
of the HP C compiler thinks it's not, you might want to report that as
a bug.
