[omniORB] omniORBpy 2.2 and AIX : problem in ORB_init()

[Thomas BARON]

Hi,
Some news from my experiments with omniORB/omniORBpy on AIX.

Well, I stop trying to build OO/OOpy using the new build system. Why ? 
First, I modify each dir.mk file in OOpy in a dirty way (copying infos 
from AIX section to the autoconf section of the file). OOpy starts to 
build but complains on a codeset symbol undefined. I know I have to  
apply the patch from Mark Zimmerman (posted on Fri Mar 21 at 17:38:02 
2003, in a mail with title "Patch for AIX traditional compile"), but I 
don't have enough time right now to figure out if this patch is 
appliable on new build method. I will try to find a little time to clean 
my changes in dir.mk and to find how to apply Mark's patch with new 
build system. I will send the appropriate patch, I think it could be 
useful for AIX users.

So I returned to old build system, and get OO and OOpy to build for my 
system. No particular problem for that.
And now, a new problem at runtime : when trying in a Python code (under 
AIX) to initialize the ORB ( using CORBA.ORB_init(sys.argv,CORBA.ORB_ID) 
), I get an exception with a message just before "omniORB: ORB_init 
failed : unknown option (id) in argument" (coming from corbaOrb.cc in 
src/lib/omniORB/orbcore, line 571). Running the same code on Linux or 
IRIX (built with new build system) causes no error at all.
But obviously, CORBA.ORB_ID is good : "omniORB4" on every platform.

Any idea about that ?

Cheers,
Tom