[omniORB] Problems building omniORB3 on SunOS 5.6

[Duncan Grisby]

On Wednesday 10 January, Richard.Turner@aculab.com wrote:

> All appears to be well until I get to here ...

[...]
> ../../../../bin/sun4_sosV_5.6/omkdepend: warning:  (from idlpython.cc)
> idlpython
> .cc: 302: #    error "omniidl requires Python 1.5.2 or higher"
> 
> This looks suspicious already as omnipython claims that it _is_ 1.5.2 ...

Don't worry about that. omkdepend doesn't understand #ifdefs properly,
so it tends to output spurious warnings. It doesn't affect anything.

> $ ./omnipython
> Could not find platform independent libraries <prefix>
[...]

The errors you get on loading omnipython are because you are running
it with a relative path. The trick it uses to find its libraries
relies on it having an absolute path, or being found in $PATH. Again,
it's nothing to worry about since the build process uses an absolute
path.

> The build continues but eventally collapses at ...
> 
> make[2]: Entering directory `/usr/local/src/richard/omni/src/lib/omniORB2'
> + mkdir -p omniORB3
> ../../../bin/sun4_sosV_5.6/omniidl -bcxx -Wba -p../../../src/lib/omniORB2
> -ComniORB3 ../../../idl/Naming.idl
> Segmentation Fault

Now that's bad. I don't know why it would segfault. Can you look at
the core file in dbx, and see where it died?  The executable for the
core is omnipython. There isn't any debugging information in either
omnipython or the omniidl shared library, but you should at least be
able to see which function it died in.

It's possible that the omnipython binary for SunOS 5.6 is broken in
some way. We don't have any 5.6 machines here, but someone told us
that the 5.5 binary worked fine on 5.6 too. One thing you can try is
to compile and install a full version of Python, from the py152.tgz
file. That might help.

Cheers,

Duncan.

-- 
 -- Duncan Grisby  \  Research Engineer  --
  -- AT&T Laboratories Cambridge          --
   -- http://www.uk.research.att.com/~dpg1 --