With the baby AS/400 DO2 machine due to be announced today – in Europe at least – IBM Corp says that it is committed to providing Posix compliance to OS/400 at some point in the next few years. According to Dave Zilles, Application Business Systems market plans and support, it will enable IBM to satisfy the criteria of a mythical checklist and to meet the needs of a limited number of users – c’mon, the only reason for it is that it will be a requirement in a growing number of US Federal Government tenders and IBM doesn’t want the AS/400 to be locked out of that market. However, he says, it will not make any significant difference to users or to IBM since customers are demanding application portability rather than some half-baked standard meant to make sense of Unix. Further, Zilles denies that Unix equals open. Users still have to select a database on which to run the plethora of Unix applications, and working with a third-party database is not dissimilar to operating under a proprietary operating system and hardware. Zilles maintains that once a layered environment is created in Unix, it is not radically different from the AS/400, and a key difference between Unix and the AS/400 is the hidden cost of an open systems approach. All of which sounds eminently reasonable many would agree that a complete Unix system does not come cheap.

Ill-defined

But, is an ill-defined checklist reason enough for IBM to provide Posix compliance? Are the needs of a limited group of users so important to IBM? If those customers are government UK, US federal and state – then the answer might be yes, but Zilles insists that IBM has won contracts where Posix was a requirement. So why do it, and if IBM is committed to Posix, why not announce when it will be available? The answer might lie in IBM’s involvement with the Open Software Foundation and its plans for the Distributed Computing Environment and Distributed Management Environment. If IBM is serious about enabling application portability between different hardware, and that is the the lynchpin of the Distributed Management Environment, the AS/400 must be Posix-compliant in order to participate. According to Ron Fess, manager of Application Business Systems Openness & Object Orientation, Posix will be an enabler to participating in the Distributed Computer Environment, an essential for interoperability since IBM cannot match its hardware on an individual basis with the other members. Further, Fess says that the DCEers will be hooked via communications so that hardware commonality almost becomes irrelevant. Nonetheless, he acknowledges that the AS/400 hardware will have to change. It is still tied into the System/38 mode, but that is under evaluation and it will evolve into some level of common hardware, and Fess says that might be based on 390, Motorola or Intel, although the first seems most improbable. As to how Posix will be implemented, a Posix/Distributed Computing Environment will sit beside the AS/400 applications and a task switch will alternate between the two environments, enabling both to access one database.

By Janice McGinn

Fess states that the strategy is to support additional standards, to enable independent software development and independent device attach, and to enable user defined objects. Given that IBM has kept its inherent object-orientated architecture a deep, dark and evidently shameful secret since the System 38 was first launched, one cannot help but wonder why it has become a talking point over the past few months. Rumour has it that forthcoming AD/Cycle announcements will feature a deal of object technology, and the Object Management Group is confident that IBM and Apple Computer Inc will adopt the Object Model, Object Request Broker and Type Repositories in their collaboration with Metaphor Computer Systems Inc. Also, the Software Foundation’s original request for technology for the Distributed Management Environment included a section called Management Options which stated that object orientation was a requirement for the submissions, and all six submiss

ions included object management. Anyway, if IBM believes that users are now adult enough to behave responsibly when exposed to the heady and dangerous brave new world of objects and that it won’t scare off the non-tecchie types, how does Fess plan open up the system and enable users to define their own objects? He says that objects will be grouped as a single entity in a relationship and that relationship might include database and display files, data and programs. This entity can then be moved to different parts of a system to be manipulated and used. The AS/400 has persistent objects as opposed to the temporary objects in SmallTalk, and Fess believes that the former has advantages in terms of single-level storage and other features that are already incorporated in the AS/400. The key is how to bind methods with data without making the AS/400 unwieldy, and Fess says that the DCE-Posix-Object Oriented Programming System enhancements would mean that the 11m lines of code would have to increase, possibly by 15%. These capabilities would probably be available as a separate licence or optionally installed feature, but Fess stresses that he is at the prototype stage at present, and it is difficult to assess how much code and additional disk storage would be required. Getting to a classical object oriented programming methodology will take time, but output from a C++ compiler that flows into a C compiler is one method under evaluation. Fess acknowledges that if the AS/400 is going to support the Distributed Computing Environment, IBM will have to find a way of supporting C.

Mid-range troubles

Since submissions for technology for the Foundation are written in C, and since Fess hopes to see the Distributed Computing and Management Environments become complementary, it can only be a only matter of time before IBM codes in C. He may only be at the prototype stage, and no doubt Ron Fess has a plethora of ideas for the AS/400 that are still under wraps. But the combination of Posix, additional application interfaces, object oriented technology, systems programming in C and the Distributed Computing and Management Environments, has to make one stop and think. Few people would agree with IBM that the AS/400 is open and IBM’s history suggests that it never will be – but the company has its fair share of mid-range troubles. Many of those are of its own making, with the RS/6000 a case in point, but the Unix vendors and Japanese are not going to go away. That particular Hydra has assumed alarming proportions with the 44% shareholding that Fujitsu holds in HaL Computer Systems Inc, established by one Andrew Heller, formerly the RS/6000 design team leader (CI No 1,748). It means that Fujitsu has its very own mid-range Unix company – in addition to ICL’s extensive technology. Necessity, usually an unpleasant compulsion, may well be the mother of IBM’s invention.