We’re told there should be little trouble merging those three submissions to the Object Management Group’s Part A Common Object Model-to-Corba Request for Proposals (CI No 2,735). Although they each cover a different scope or area of mapping object models, they are fairly similar. As Visual Edge Technology Ltd’s Objectmap is more complete, it is likely to take centre stage – with gradual modifications as the merged submission is formed. The way we hear it, Fujitsu Ltd’s mapping focuses on describing the different possible implementation choices for performing an Object Linking & Embedding Automation to Corba mapping, but does not provide significant details about the mapping itself. The Sybase-Genesis-Expersoft mapping also covers only the Object Linking & Embedding-to-Corba mapping, mostly the type mapping issues. It does not address many of the issues relating to creating or activating objects across system boundaries. It also does not cover important issues such as object lifecycle. Visual Edge’s Objectmap covers the mapping of both Common Object Model and Object Linking & Embedding to Corba. It does not attempt to describe the different implementation choices, though it does deal with the issues of implementation incompatibilities between Common Object Model on Win16 and Common Object Model on Win32. Digital Equipment Corp’s submission, left out of this single-submission get-together, maps MIDL (the description language used for Common Object Model on Win32) to Corba Interface Definition Language. It also provides a mapping for Object Linking & Embedding. It provides little detail about lifecycle, creation, activation and inheritance mappings and tends to have a bias towards an implementation based on Distributed Computing Environment. One player admits the technical issues in merging with the DEC submission are not the real problem. The problems are political and revolve around the same issues that surfaced during the UNO versus Di stributed Computing Environment fight. Each of the submissions has parties that are strongly religious about one or the other. However, to produce a mapping from Common Object Model to Corba, the wire protocol (if any) is, in reality, irrelevant since it never appears on the surface to the user of an interoperability system. User needs will drive which wire protocol is chosen in any given situation.
