Why IDLs are not ideal

The dominant approach to addressing heterogeneity, interoperability and legacy software components at present is based on the use of interface description languages (IDLs) such as the OMG/CORBA IDL. We believe that this approach has serious drawbacks. In this paper we outline our objections to the IDL-based approach, then describe ongoing research directed toward producing a superior alternative, which we refer to as the polylingual systems approach. We illustrate both our objections to the IDL-based approach and also our new polylingual systems approach with examples based on the IWSSD common case study.