At present, the push to get the Open Software Foundation’s Architecture-Neutral Distribution Format is stalled, but X/Open Co Ltd’s European Operations manager Paul Tanner argues that if computer users are convinced that adherence to a standard will give them a clear commercial benefit, they will make it a requirement, thereby putting a value on it, such demand will justify the necessary investment by suppliers to meet the remaining conditions (CI No 2,339). As of now, the Architecture-Neutral Distribution Format is clearly not supported by many systems suppliers or independent software vendors. To discern from where support may come, the Distribution Format must be looked at in a different way, says Tanner.
Make process
The idea is that application developers could begin to use the UK Defence Research Agency’s TenDRA technology as a means of checking adherence to application programming interfaces long before they adopt the Architecture-Neutral Distribution Format as a distribution product. Indeed, all ANDF proponents concur that TenDRA offers significant benefits that result from the enforcement of correct use of application programming interface sets. It works like this, says Tanner. In the normal process of making (compiling, linking and installing) an application, it is current practice to make numerous assumptions about the interface to the target system. Functional testing is used to check these assumptions before the software is shipped. However, the assumptions made by the system provider may be subtly different, possibly with serious consequences. Using the TenDRA approach, a much greater degree of testing is introduced by precisely informing the make process about the application programming interface set that is being used. According to the Architecture-Neutral Distribution Format proponents, this approach is all the more valuable in view of the fact that there is no single application programming interface view of the world, rather it represents an evolution towards an ideal. Add to this the consideration that the target system will often be pieced together from the offerings of multiple vendors and there are compelling reasons to check that the same assumptions are being made. Despite X/Open’s best efforts at standardisation under initiatives such as the Common Application Environment, X/Open’s Tanner says that the independent software vendors it has polled complain that it is still costing them too much to convert to and support multiple systems. In which case, he argues, independent software vendors, using the methods described above, could use TenDRA technology to assist these efforts long before they adopt the full Architecture-Neutral Distribution Format.
By William Fellows
It should also create the conditions for a gradual increase in the use of the distribution format. This means that the Architecture-Neutral Distribution Format’s chances of success are much greater than they would be – it will take a while to happen, but there is no reason to give up hope just yet. Apart from the Open Software Foundation and Novell Inc efforts, the European Commission’s Open Microprocessor Initiative has been supporting the Distribution Format for some time, hoping that it will facilitate the creation of an environment in which new low-level technologies can be introduced into the market on an affordable scale. An initial project funded the development of the Architecture-Neutral Distribution Format installers for several leading systems. A second phase, known as the Deploy project, approved by the European Commission just before Christmas, should be under way soon. Project members, apart from X/Open, are the Defence Research Agency, the Open Software Foundation Research Institute, IXI Ltd, Software AG and two consultancies, Etnoteam and ET International. As well as providing application programming interfaces as a full partner in the project, Tanner hopes X/Open’s independent software vendor and user members will give the organisation the go-ahead to spend money to get involved as an active particip
ant and stimulate the development of an application testing capability. X/Open is already working in conjunction with the Japan’s Ministry of International Trade & Industry on a generic project to specify an interface using existing technologies from Sun Labs. This kind of effort could open up a market for a range of testing tools that use the Distribution Format mechanisms as their base, says Tanner. If X/Open gets the nod from its members – and it will get the first indicators in two or three months’ time, Tanner says – it will begin to look at ways of providing TenDRA-based constructs for testing components such as CDE and Spec 1170. (Before this can happen, Tanner says that the Spec 1170 application programming interface must first be more clearly defined. He says X/Open will likely do what the vendors did in the first place to put it together, such as asking independent software vendors, developers and suppliers just which parts of Unix application programming interface sets they actually make use of). Tanner hopes that interest in ANDF-based application programming interface testing techniques from other industry organisations, such as the Object Management Group, can be turned into concrete support.
Workload
The Object Group, like many other standards bodies, is heaving under its workload and is looking to borrow proven conversion, testing and verification technologies where it can rather than doing the job itself from scratch. X/Open’s interest in object paradigms will in any case lag sometime behind the Object Group’s frontier. Tanner says object techniques must first pass from the expert domain into more general usage before X/Open can begin to find out what kinds of object standards are needed. Pressed into use in these kinds of services, the Distribution Format has a window of opportunity, and Tanner believes ANDF can prevail if independent software vendors support the current initiative. If they don’t, success may still be possible but I, for one, will not be betting on it, he warns.
