The European Xbase Institute
Institut Xbase Européen
Issue Nr 1 - 1992
by François Ghoche
May 6, 1992, will remain in our memory as the birth of an Xbase standard. This is the day the standard proposal passed its first official milestone.
It will take some time before Xbase takes it to its final shape, but the first step is done. This is, in itself, a success, considering the major obstacles that have been raised in front of all former trials since 1987.
For the time being, the initiative is internal to the United States, the project being submitted to ANSI (1). Nevertheless, it is necessary to foresee from the beginning the extension of the standard to the international level. By doing so, we shall avoid facing the same problems as have been already ocurred with other standards in the past.
We have already contacted the initiators of the project in the US. They approve the idea of including an "international" item in the agenda of the normalization committee. To be sure to take into account the greatest number of qualified opinions, it is necessary that all those concerned express their ideas as soon as possible. The first meeting of the ANSI Committee will probably take place sometime in Autumn. Our proposals should have been submitted by that time. Meanwhile, we shall develop contacts with correspondents from Japan, as well as European countries, to try build a coherent project. It will then be taken into account from the start of the process.
A meeting will be organized soon to facilitate the exchange of opinions and to spread the information on hand. If you wish to come to that meeting, please get in touch with us. We shall keep you informed as soon as possible.
The proposal for an Xbase standard was presented by Marc Schnapp, Martin Rinehart and Bill House. The proposal has been presented to the committee in charge of the examination of the normalization projects, the SPARC (2) committee of X3. Let us recall briefly that X3 is a US standard development organization accredited by ANSI. It enjoys the support of software and hardware vendors, as well as customer organizations. The X3 secretariat is handled by CBEMA (3).
SPARC proceeds with the preliminary study of the standard proposals. A proposal accepted by SPARC must then be confirmed by a ballot involving all the members of X3. As far as Xbase is concerned, the result of the vote should be known sometime in July. The outcome should be positive, as the SPARC advice is usually followed.
The former attempt to standardize on dBASE was launched in 1987 by Martin Rinehart. One year later, Mr Rinehart formed an IEEE committee. But after the first meeting of the committee sponsored by IEEE was held, Ashton-Tate lawyers stopped the process, arguing that everything coming from dBASE was proprietary, including the language, yet inspired from JPLDIS (4).
Ashton-Tate had subsequently formed its own structure, that included consultants from outside the company : the dLAB. It was intended to act as an advisory board on the future of dBASE IV.
The legal clouds hung over Xbase went on until the Ashton-Tate buyout by Borland International, last summer. The American Department of Justice tied its approval of the merger to an agreement by Borland to abandon all claims on elements of the language or the user interface menus for 10 years.
Since these events, the movement for an Xbase standard has been revived. A debate was engaged, and it has been going along ever since on networks such as CompuServe, and within user groups. In France, La Lettre de dBASE launched a campaign for the establishing of a standard, as well as a survey among its readers.
During the last few months, events accelerated. After the merger between Ashton-Tate and Borland, last year, we have subsequently witnessed the buyout of Fox Software -- publisher of FoxPro -- by Microsoft, and of Nantucket -- publisher of Clipper -- by Computer Associates.
The main Xbase products are now in the hands of the three leaders of the software market. These giants are preparing their weapons for the battles that are to begin sometime during the second half of this year.
Meanwhile, an active group gathered around the standard project. Multiple contacts were made with the main publishers of Xbase products, as well as third party vendors (utilities, design tools, code and application generators). We can say they have been welcomed in every case. The real question was to know who was going to make the first step.
However, these events helped to free the road for those in favor of a standard officially drawn under the aegis of an ad hoc organization. We think they have been encouraged by the main publishers concerned. As a confirmation of this analysis, we can consider the first professional event to take place, to our knowledge, after the proposal was made.
On May 12, an MMA (5) meeting was held in New York. The panel included key people from Fox Software, Borland International and Computer Associates. From what came to us about this meeting, it seems the participants were almost unanimous about the standard project. For instance, Walt Kennamer (Fox) and Mike Masterson (Borland) expressed their views that an Xbase standard would be a positive point for their clients as well as for their companies. Anders Vinberg (Computer Associates) would have expressed that his company usually approves standards.
Two key people of the New York Professional Database Association, as well as numerous members of this association, were present at the meeting and they participated in the debate. The general feeling was that the standardization move had started with a positive momentum.
On the top of the list of the questions asked was the use of different Xbase products on the same network. All professionals aware of the dBASE and compatible products currently on the market will appreciate...
This kind of question is a good example of the problems which developers and microcomputers' managers currently have to face.
The original subject of the debate was about how to choose the adequate Xbase product for the job. The subject helped raising one of the main reasons for a standard. In fact, every Xbase product on the market has strong points and specific capabilities. It seems judicious to choose the right product for every job.
As long as you only need one product for a specific type of application, the reasoning is correct. But if you have to deal with different applications, and further, if these applications have to live together on a network, we have to admit that the most reasonable solution today is to choose one and only one product, and stick with it. This way, one can avoid differences between language syntax, the learning curve of different user interfaces and different development tools, incompatible locks on networks, etc.
The current project will try to solve these kinds of difficulties. The standard project's primary objective will be to define the Xbase programming language. As a result, user interfaces and development tools are not part of the discussion. The underlying idea is that one of the main strengths of the Xbase market resides in the language it inherits from dBASE.
Of course, we know the future belongs to CASE tools. But developers know that whatever the capacities of application or object generators, these are never universal. Furthermore, sophisticated tools often contain minor defaults. The ability to examine and eventually modify the result of a robot's work is of great importance, at least when the tool is not just intended for the individual use of the end user.
Languages close to the machine, such as assembler, of course, and even C, have to be left aside when business applications are concerned. These languages require a much more important development effort, which is usually not justified for applications not made for hundreds of users. Xbase contains 4GL features of data manipulation language handling access to a database engine, which is far from being the case for third generation languages. The latter require more programming effort, even if supplementary libraries are being used.
Further, the Xbase language has the prerequisites necessary to become the standard for a database application programming language. It is taught in numerous programming cursus, it is currently known by a large number of programmers, and it is curently implemented in a great number of applications being used in a large spectrum of economical activities. To save the investment that constitutes these million lines of programs written in dBASE or one of its derivatives is in itself good reason enough for a normalization effort. Nevertheless, the usage of these languages carries some drawbacks as well. Which of these can we face with a standard ?
Precise description of the commands' syntax within the standard will enable easier migration of applications between environments. Currently, the language contains significant differences, from one product to the other. Further, learning a standard language in the different educational institutions will provide the industry with programmers immediately productive. The only training required will then concern the different CASE tools proposed by the different vendors, which competitive features will keep on the richness and diversity of the market : normalization does not mean uniformity.
The process will probably require many steps. But even if the first level standardizes on functionalities traditional to DOS releases of Xbase products, we will very rapidly have to consider events programming, GUI features, and increasing object orientation of products and tools.
It is important as well to envision, if not absolute standardization, at least publishing of network locking algorithms, as well as index file formats. It will then be possible for third party vendors to produce tools enabling different applications sharing the same data to operate on a network.
Further, the existence of a standard will encourage the creation of Xbase database servers, accepting queries issued by different products. The data dictionary concept will eventually have to be faced sooner or later.
Of course, the servers will have to be able to handle concurrently record oriented queries, made in dBASE like language, and queries on sets, made by way of languages such as SQL. It remains to be decided if a subset of SQL has to be covered by the standard. Still, the concurrent use of both techniques will be covered in future products to come, such as Borland's InterBase engine, for instance.
After this quick overview, we now have to prepare the road for the birth of the Xbase standard. We shall work for that in common, on both sides of the Atlantic.
François Ghoche - Summer 1992
Possible choices / Choix possibles :
* (© copyright 1992-1996, François Ghoche)
fg - dernière MàJ/last update: 19 Feb 96