The article they didn’t let you read

Following Father Bertie Park’s auspicious debut as a published author, I can now reveal that I myself enjoyed a brief ‘career’ as an info-tech writer in the early ’90s.

I contributed, unpaid, several articles to the Melbourne PC User Group’s organ, PC Update. Most notable among these was an article in the August 1992 edition, “Paradox 3.5 — Criticisms, Quibbles and Cautions.”

This article advanced the thesis that Borland’s relational database software product, Paradox, did indeed present, as their blurb contended, “a most ingenious paradox”: How can something so mind-bogglingly useful and powerful be such a dog to use?

The magazine ran a response from Borland International (Australia) — which, as it happens, was a patron of MPCUG — directly following my article:

Borland’s Response

Mr Stam’s feelings about the ageing Lotus-style menu system in Paradox 3.5 were shared by many users, which is why Borland replaced it in Paradox 4.0 with a new-generation TurboVision-style interface, complete with mouse control, pull-down menus, dialogue boxes, radio buttons and so on. It now shares a common look and feel to Borland DOS-based language and spreadsheet products, making it much simpler to move back and forth between applications.

His various points are well taken, and some of them have been addressed in Paradox 4.0, while others are more related to philosophical differences between users. We try to address these individual preferences where possible — for example, Paradox 4.0 allows diehard “ParaLotus” menu fans to switch back to the older menus if required.

Paradox 4.0 has been independently benchmarked as the fastest multi-user PC relational database on the planet, as PC Update readers are probably aware of by now. Melbourne has always been a hotbed of Paradox activity, and we welcome further feedback from Mr Stam or any other readers. In fact, Borland Australia selected the Melbourne leg of our recent Borland 92 technology seminar as the official launching pad of Paradox 4.0 in Australia. We look forward to having the opportunity of demonstrating Paradox 4.0, along with many other new Borland products, to a future Melb PC meeting.

Ian Robinson
Marketing Manager

Whetted with Mr Robinson’s invitation to contribute “further feedback”, I boiled down my impressions on Paradox 4.0 in a further article. PC Update said they would like to use it, but could I please cut it down to 1000 words or so?

Well, I did, but in the event the magazine did not publish. So, here, for the first time is my full, original article on Paradox 4.0 — the article they didn’t let you read!

Paradox 4.0 — Quibbler’s Response

An article of mine appeared in the August edition of PC Update which was somewhat critical of various aspects of the now superseded version 3.5 of Borland’s Paradox. Tacked to the end of it was a short item headed “Borland’s Response”, which was a rebuttal from Mr Ian Robinson, Borland’s Marketing Manager here in Australia. Mr Robinson wanted to assure readers that some of the less pleasing aspects of the program I had identified were to be addressed in Paradox 4.0, the release of which was quite imminent.

My criticism focussed in particular on what I saw as Paradox’s antiquated and cumbersome user interface. Readers may recall the example I gave of the tortuous means by which report settings are fixed in Report Mode. This type of thing, Mr Robinson suggested, was to be remedied in Paradox 4 with the introduction of mouse support and a new “Windows-like” interface.

After spending some time exploring Paradox 4, I have to say that my response is lukewarm as far as the user interface enhancements are concerned. I was disappointed to find the very same sinuous user interface, albeit smodernised with drop-down menus, pop-up windows, list-boxes, etc.

In short, the version 4 enhancements to the Paradox front-end are largely cosmetic. The main difference seems to be that, rather than having all the interaction take place on the two top lines of the screen, the user is treated to something like a pyrotechnic display as more menus, input boxes, etc., erupt all across the workspace. This may or may not be an improvement, depending on one’s aesthetic sense, but also on practical considerations such as keeping what’s on the workspace in plain view.

Although Paradox’s mouse support can sometimes be helpful, one is nevertheless stuck with that tortuous menu structure which was my chief concern in the previous article. To hark back to the example I gave there, the Report Mode menu structure must be traversed repeatedly for each and every report setting one wishes to modify. My vision, in the June article, of a “Report Settings Dialogue Box”, in which all the report settings are available for perusal and/or modification, was not to be.

Mouse support, inexplicably, is simply not there in certain situations. For instance, after selecting Field/WordWrap from the Report Mode menu, one cannot use the mouse to click-select the field one wishes to wordwrap. Paradox simply beeps stupidly, obstinately waiting for one to instead use the cursor keys to place the cursor on the desired field. One then presses the Enter key to select the field. Forget the mouse for the moment, because Paradox already has. Another disappointment was that the mouse cannot be used to “drag” a field to another location in a report spec.

The conclusion I’ve reached regarding the Paradox user interface is that it cannot be other than what it is — that is, labyrinthine and cumbersome. To change the structure of the user interface would mean making it incompatible with an established user-base whose existing Paradox Application Language (PAL) applications depend on that structure. In order not to estrange itself from its user-base, Paradox must maintain the status quo, because a PAL script is just that — a script — meaning, it simulates user interaction within a presupposed structure. The effect of this is familiar to most PAL programmers, who know what happens when a script tries to select a non-existent menu item.

The instructions in a PAL script have the character of being “interface-based”, which is in marked distinction to a truly command-based programming language such as used in dBase. Even some PAL instructions which appear to be “commands” (such as VIEW, EDIT, SORT, etc.) are really only shorthand for the corresponding menu interactions. Over the years, of course, PAL has been given extensions which obviate its more serious deficiencies, some of which are genuine “commands”. Some indeed offer functionality not available in Paradox’s Interactive Mode.

However, the fact remains that Paradox, as a development environment, is seriously hide-bound. In order to become a “true” programming language, PAL requires a ground-up overhaul, with extensions upon extensions. Only when PAL has evolved to a point where scripts can run independently of the interface structure will Borland be free to give Paradox the user interface it needs and deserves. Only then will Paradox truly be a pleasure to use interactively. And only then will Paradox truly be a pleasure to program.

A problem here is that established Paradox users are justifiably wary of changes which would nullify the considerable time and effort invested in their existing systems. Like so many aspects of technology, PAL as it exists has become — and probably will continue to be — entrenched not because it is ideal, but because users have made do with it for so long that they are accustomed to it — and dependent upon it. Many, I suspect, are even genuinely fond of it.

I mentioned in the June article my annoyance with the Paradox 3.5 feature in which the only way to save a report, form or script on which one is working is to press the Do_It! key (F2). This has the added effect of exiting back to Main Mode, and tough-luck if one wanted to continue designing one’s report. Okay, so it’s not a severe flaw, but I thought a worthwhile enhancement might be to provide an option which would allow one to save the work done so far and continue working uninterrupted. I was mildly disappointed to find that Paradox 4 has not improved upon this time-wasting “design feature”.

To take up another matter raised in my June article, I regret to report that the “CalcEdit bug” discussed therein has not been corrected in Paradox 4. It still lurks in Report Mode to “entrap the unwary”. And the fact that Borland have omitted to warn users in the documentation, even though they are aware of the problem, ensures that there will always be the odd unwary user to be entrapped.

I shouldn’t neglect to report on some of the more pleasing aspects of the Paradox 4 upgrade. Not all the enhancements to the user interface are merely cosmetic. The “window-based” interface allows easy manipulation of images on the workspace. One can move, zoom, re-size or close windows, adjust table column size and (in most circumstances) select objects with a few deft mouse actions. The look-and-feel of the interface might be familiar to some as the Turbo Vision type used extensively in other Borland, particularly language, products.

PAL programmers are at last blessed with a half-way decent script editor. The script editor is, in fact, very much like the editors of Borland’s Turbo languages, with mouse support, search and replace (exact match or pattern), block cut and paste — in short, all the basic features one would expect in a text editor from a major software company.

Of course, there are enhancements beneath the surface, too. Execution of queries is, we are assured, significantly accelerated in comparison to previous versions. Borland have also given users a way of viewing a table in an order other than that in which it is keyed, by means of secondary indexes. It is perhaps odd that it took 4 versions for what should be a rather basic feature in a major relational database package to become available for Paradox users. Well, it took Borland to do it...

By far the most exciting enhancements for many will be the PAL extensions, which number some 100 or so. These provide, among other things, event-driven programming capabilities, and generation of Windows-like user interface objects, such as windows, dialogue boxes, etc. Used skilfully, they can greatly enhance the effectiveness of user interaction within PAL applications, as well as providing a bit of pizzazz.

I suppose no software product will ever be perfect. To put that another way, no software product can ever be all things to all users. All the same, Paradox 4 is a significant and worthwhile upgrade. One might reckon the upgrade price of $295 (from v3.5) to be rather steep until one considers the amount of development work which must have gone into the product.