Re: Tn5250 compile under IBM Aix 5.3

There was a discussion on another list I'm subscribed to a few days ago
about the differences in coding for AIX and other non-win32 operating
systems. One of the big things is that, apparently, you have to make some
changes to the defines and undefines for those O/Ses. Might want to check
and see if anyone has *ever* compiled a version of VNC for AIX.




-----Original Message-----
From: linux5250-bounces@xxxxxxxxxxxx [mailto:linux5250-bounces@xxxxxxxxxxxx]
On Behalf Of Alberto
Sent: Wednesday, May 12, 2010 11:28 AM
To: linux5250@xxxxxxxxxxxx
Subject: Re: [LINUX5250] Tn5250 compile under IBM Aix 5.3

Hi James, thanks for your answer :-)
I succesfully compiled the tn5250 under aix at least, but unfortunately
every time i try to launch it i get a "segmentation fault(coredump)" error.
The compiler was XLC , seems that with GCC there still are problems.
I will try to do some tests and i'll let you know :-)
Thanks.
Ciao
Alberto.

"James Rich" <james@xxxxxxxxxxx> ha scritto nel messaggio
news:mailman.26695.1273172180.2580.linux5250@xxxxxxxxxxxxxxx

On Thu, 6 May 2010, Carey Evans wrote:

AIX redefines accept to naccept in sys/socket.h:

#define recvfrom nrecvfrom
#define accept naccept
#define getpeername ngetpeername
#define getsockname ngetsockname

which means that line 466 of sslstream.c ends up trying to assign to
the naccept member of the struct. You could try using the C++ compiler
instead, which uses an inline function instead of #define, or simply
add #undef accept near the top of the file.

I've been searching for a way to handle this in autoconf. Surely we can't
be the only ones who have this problem. I haven't found anything useful
so far.

Alberto, I think you need the #undef at the top of sslstream.c.

James Rich

if you want to understand why that is, there are many good books on
the design of operating systems. please pass them along to redmond
when you're done reading them :)
- Paul Davis on ardour-dev