[omniORB] regression bug: omniORB4.0 CVS snapshot 20031022 compile failure on RH Linux 7.3

[Bastiaan Bakker]

Hi,

I'm trying to build omniORB4.0 CVS snapshot 20031022. On RH Linux 9 it
compiles without problems, but on RH 7.3 the build fails with the
following error message:

omniidl: Preprocessing '../../../idl/Naming.idl' with
'/home/lmsbuild/sources/cvs.lifeline.dev/lifeline/rpm/BUILD/omni/lib/omnicpp -lang-c++ -undef -D__OMNIIDL__=0x2420 -D__OMNIIDL_CXX__ "../../../idl/Naming.idl"'
omniidl: Running front end
omniidl: Running back-end 'cxx'
omniidl: fatalError occurred, in debug mode.
>> An internal exception was caught
Stack:
-------------------------
  File "../../../bin/omniidl", line 105, in ?
  File "./main.py", line 486, in main
  File "../../../src/lib/omniORB/omniidl_be/cxx/__init__.py", line 306,
in run
  File "../../../src/lib/omniORB/omniidl_be/cxx/util.py", line 158, in
fatalError
Exception:
-------------------------
Traceback (innermost last):
  File "../../../src/lib/omniORB/omniidl_be/cxx/__init__.py", line 266,
in run
  File "../../../src/lib/omniORB/omniidl_be/cxx/descriptor.py", line 87,
in __init__
  File "./idlast.py", line 171, in accept
  File "../../../src/lib/omniORB/omniidl_be/cxx/descriptor.py", line
116, in visitAST
  File "./idlast.py", line 300, in accept
  File "../../../src/lib/omniORB/omniidl_be/cxx/descriptor.py", line
122, in visitModule
  File "./idlast.py", line 345, in accept
  File "../../../src/lib/omniORB/omniidl_be/cxx/descriptor.py", line
131, in visitInterface
  File "../../../src/lib/omniORB/omniidl_be/cxx/descriptor.py", line
168, in initialise_base
  File "../../../src/lib/omniORB/omniidl_be/cxx/descriptor.py", line
221, in hex_word
OverflowError: long int too long to convert
make[4]: *** [omniORB4/Naming.hh] Error 1

The problem appears to be introduced in revision 1.1.4.5 of
src/lib/omniORB/omniidl_be/cxx/Attic/descriptor.py, which may help to
reduce warnings in Python 2.3 but breaks in Python 1.5.2, the release RH
7.3 ships with. At least reverting to 1.1.4.4 resolved the problem.
I don't know if it's possible to make both Python 1.5.2 and 2.3 happy.
If not, could the fix for 2.3 be pushed to the omniORB 4.1 branch ? I
think it would be a pity if such a minor issue would break RH7.3, still
a widely in use platform.

Cheers,

Bastiaan Bakker
LifeLine Networks bv