From a sister publication, Unix News International
The notion that there might be a fifth-generation language fills one with dread, if only because of the prospect of learning a new dictionary full of associated acronyms and buzzwords. Fortunately, the future of 4GLs is more about integrating existing and newly emerging technologies, rather than having to bet your career on any dramatic new developments. Specifically, there is a small handful of main considerations when looking at the future of 4GLs, or applications development generally. They are, in no particular order. Automatic code generation is no longer rocket science, and vendors are being forced to add more functionality, and links to other technologies, to maintain some level of product differentiation. Internet enablement. As we write, vendors are announcing, or preparing to announce, additions to their core development products which seamlessly integrate Web authoring and integration facilities, aimed at both external parties and internal users – the so-called Intranet. Java and Netscape are the main target technologies. Increased management facilities for the development process itself, the deployment of new applications, and for newly implemented applications. Hooks to database, application, system and network management systems are increasingly being deployed in the development stage. Components, or object-based functions, will play a growing role in development environments, ranging from in- house modules of useful legacy code, to full-blown application components (or applets), which may be as large as a sales ledger operation. Integrating components, and applying them specifically to business requirements, rather than developing them from scratch, will become the key areas of focus. Depending on which supplier you speak to, one of these points will be their main area of development and marketing focus. Informix, for instance, has a heavy bias towards a component-based approach with New Era, while Sybase, via Powerbuilder, is spending much effort on the Internet and Web fronts. Supplier focus also depends greatly on where it has come from. IBM, naturally enough, brings with it a heavy emphasis on enterprise development and management, as do the high-end specialists Seer and Forte, and the perennial enterprise supplier, Texas Instruments, formerly IEF. At the mid- to high-end, suppliers such as Unify, Progress and Uniface have been quick to marry PC-based visual development tools with rapid application development (RAD) technologies and enterprise management facilities, as have the traditional midrange targeters such as Cognos, Synon and MDIS.
Degrees of success
From the low tools perspective, Microsoft, Powerbuilder and Borland were quick off the mark at the graphical PC stage, and are now trying to work their way up to the middle- and high-end enterprise environment, with varying degrees of success. All these technologies are trying desperately to find a happy middle ground despite the different historical approaches. The main question, in terms of supplier success, is whether that middle point ends up more towards the high- or low-ends. In terms of the influencing technologies themselves, however, they are quite clear. While the Internet and Intranet are getting the most airplay for various fashion reasons, the technology also overlaps with management issues in a big way. The reasons are simple: when enterprise applications are accessed as Web sites, they are, by definition, easier to control. Fitting users with Web browsers immediately removes the need for large-scale version upgrades to hundreds or thousands of desktop machines, since that is automatically out when the Web site is accessed. And Web sites, while fronting potentially vast databases, are easier to manage than the databases themselves. With cost savings and efficiency gains already proving immense at pioneering sites like Federal Express, it’s little wonder the 4GL suppliers want to get in on the act.
Management tools
However, one of the p
roblems with client/server development is the large numbers of code that are created at the same time- therefore it is very important to incorporate project management, configuration management and so on, so as to build a deployment environment. Software that allows individuals to create code will always be required, but today’s trends point towards an environment in which those tools are used more for building clever hooks and links between business functions, than for the applications themselves. The shift towards a greater use of objects and/or components, points clearly to a time when application development becomes simply a matter of drawing lines on a screen between functions, business units, and company workflow rules. This has always been suggested with each new generation of languages, but now it actually looks probable, not just possible.
Content from our partners
Mythical all-rounder
The coder’s skill will be geared more towards validating those integrated components, and about control and connectivity issues, than actually generating the code. The end result is an IT- literate business person with responsibility for vertical units within the organization. But the problem with that theory is that it smacks of the mythical, hybrid IT-business manager concept. This was very popular five years ago, but interest in this idea has waned in the face of the more sensible approach of business and IT specialists working more closely together. Benefits from this are at their greatest when (RAD) tools are used in conjunction with input from all relevant parties: end-users, business managers, IT technologists and IT management. RAD may be the closest there is to mainstream programming with pictures, but the consensus remains that RAD’s usefulness as a development tool for prototypes ends there. Real code, based around the prototype’s core functions, must be developed from scratch, which brings us back to the future of 4GLs. Since business modeling must come first, the trick is being able to feed those models directly to the programming tools, regardless of brand. This balances the emphasis between developing the application in the first place, and then delivering an application which meets business requirements. Company departments have had fun playing with Visual Basic and Powerbuilder, but once the applications were built they had to turn to the IT department to manage it. But it’s the RAD approach that will help concentrate minds on modeling the business requirements.
Bane and savior
This may sound obvious, but it nevertheless remains a key issue within the application development process. As does the future of the three-tier client/server architecture, which has been first the savior and, more recently, the bane of users worldwide. The advent of the Intranet, which can be designed for ultra-thin clients, takes away much of the focus on partitioned applications, which have been heavily promoted by vendors of the so-called second-generation 4GLs. The real payback in client/server is not departmental systems, but enterprise systems – that is, alternatives to mainframes. The resurgence of investment in mainframes over the last 18 months is because companies have been getting the performance, scalability and manageability from client/ server. And that’s because the 4GLs people have been using simply aren’t up to it. Or, perhaps more precisely, the current crop of 4GLs have not been used in conjunction with the necessary components and good development practices essential to large-scale deployment.
INTERNATIONAL
