[omniORB] Does omniORBpy support long double yet?

Harri Pasanen harri.pasanen@trema.com
Thu, 05 Apr 2001 10:40:14 +0200


Craig Rodrigues wrote:
> 
> On Thu, Apr 05, 2001 at 01:45:08PM +1000, Stephen Crawley wrote:
> >
> > > I am trying to compile some 3rd party IDL that contains
> > > all sorts of wacky IDL types like "long double".
> >
> > Beware of 'long double'.  It is not currently supported by the IDL ->
> > Java mapping.  I know you are trying to use 'long double' from Python
> > ... but if you care about interoperability with Java, you should avoid
> > using 'long double'.
> 
> I don't care about interoperability with Java (yet).
> I am interested in compiling some third party IDL files
> available here:
> http://www.jtrs.sarda.army.mil/docs/documents/sca.html
> 
> These files include wchar, and long double.
> 
> I'm not even using the interfaces which use wchar and long
> double.
> 
> I can hack these IDL files so that they don't include
> wchar and long double, but I'd rather not.
> 
> Is there an option I can pass to omniidl, so that it will
> not choke on an IDL file, but instead will refuse to generate
> code for types it does not support?
> 

A pragmatic solution:

It should be a straight forward exercise to write a preprocessor for
omniidl in Python that would strip out those interfaces with unsupported
types.

A similiar situation arises with yacc and flex generated code that must
be compiled with ANSI C++ compliant compiler, in this case we have to
post process the generated source.

Harri