[omniORB] AIX / Visual Age 5 / omniOrb 3.0.2

[Gary D. Duzan]

In Message <20010430124848.A48362@frii.com> ,
   Mark Zimmerman <markzimm@frii.com> wrote:

=>On Mon, Apr 30, 2001 at 01:08:45PM -0400, Gary D. Duzan wrote:
=>>    Check your /usr/local/include/python2.0/config.h file and see
=>> if it has "#define inline __inline" or something of the sort in
=>> it. If so, you need to put an "#ifndef __cplusplus" / "#endif"
=>> around it. I think that should take care of most of the issues for
=>> omniidl, but there are some more changes to Python required to get
=>> it working with omniORBpy under AIX+VACPP5. I'm working on a set
=>> of patches now, and I'll probably check out the new Python 2.1
=>> release to see how it does.
=>> 
=>
=>Thanks, that got me a bit farther, but only until:
=>
=>xlC_r -c  -qstaticinline -qmaxmem=8192  -D_REENTRANT -D_THREAD_SAFE -I. -I./.. -
=>I./../.. -DUSE_omniORB_logStream -D_OMNIORB_LIBRARY -DUnixArchitecture -DCONFIG_
=>DEFAULT_LOCATION='"/etc/omniORB.cfg"' -I.  -I../../../../include -D__aix__ -D__p
=>owerpc__ -D__OSVERSION__=4 -o portableserver.o portableserver.cc
=>"./../omniORB3/ir_operators.hh", line 6.3: 1540-0218 (S) The call does not match
=>any parameter list for "::operator>>=".
=>"../../../../include/omniORB3/corba_operators.h", line 41.1: 1540-1283 (I) "oper
=>ator>>=(const CORBA::Any &, CORBA::UNKNOWN *&)" is not a viable candidate.
=>
=>
=>Do you think I might be better off with python 1.5.2?

   No, Python 2.0 is fine, and this doesn't look like a Python-related
error. You need to patch include/omniORB3/bufferedStream.h to get the
operators properly scoped. The patch is under patches/aix_xlC_5.0.patch
in CVS or the 3.0.3 release and should apply fine to 3.0.2 with only a
few lines of fuzz, or just grab the new bufferedStream.h and patch it.
The change is fairly significant and the core folks didn't want to make
it without knowing the impact to all the other ports, so the patch was
included instead.

					Gary Duzan
					Verizon IT



p.s. Python 2.1 seems to have some of the same configuration problems
w.r.t. omniORBpy as 2.0, but it looks like the inline problem has
gone away. I'll see if I can get it working and feed the config patches
back to the Python folks for a planned bug fix release.