In 1969 a landmark NATO conference was held in Europe to start a debate on a strange, new imaginary thing: software engineering. Now a familiar term, at the time the conjunction of the two words seemed as unlikely as slimmer’s beer or kosher ham. The radical notion was that software development could progress from just being a process (which it wasn’t, then) to being scientific, in the sense that instead of building systems by seat-of-the-pants blind flying they should be as well designed and robust as other engineering endeavors, such as bridge building. Here we are nearly thirty years later, and that goal is as remote as ever. The principal proof is of course the Year 2000 software crisis, caused in part by the very fact that so much code is not so much machine-tooled as hand-loomed. Software engineering caught on, somewhat, in the more technical end of the development community – hence Ada, the more rigorous Computer Aided Systems Engineering (CASE) toolsets, certain parts of the AI world (as in its use of some mathematically-based formal methods like Z, for example) – but never, ever lit the fires of the average commercial Data Processing dude. This individual was much more likely to treat such discipline as some of us do flossing our teeth – you know you should, but when do you ever have the time? The emphasis in the average DP shop was, and is, much more on the old journalistic adage, Don’t get it right – get it written, as in, strive for a deliverable, not perfection.
By Gary Flood
So it was with almost a tinge of nostalgia that we heard from Lexington, Massachusetts based Software Emancipation Technology Inc, which is pushing its Discover Development Information System. This software is claimed to parse and identify vital elements of code, then build it into a complete Information Model of the corporate software system to help organize, automate and document the software engineering process in an unprecedented manner. It’s that word automate that sticks out. How can anyone seriously claim to be able to do that in this day and age, when we have seen so many instant panaceas come and go? And given that the company slogan is from craftsmanship to engineering, we seem to doing a bit of a Back To The Future with these guys. After all, 4GLs were supposed to fix the development problem back in the 1970s, CASE came – and then went ‘phut,’ C++ was the magic bridge between machine code and OO, and now we have Java, the answer to a maiden’s portability prayers (hmmm). For its part, Software Emancipation is adamant that it’s not just the tecchies who are turned on to this stuff, but their managers, finally. The company’s roots go back to 1991, when it was founded by former Prime Computer Inc executive and Russian emigre Vladimir Geisberg, who has a heavy CAD/CAM background (which figures). We’ve tracked the company somewhat the past couple of years, and reading between the lines it seems to have been through something of a fallow patch.
Revamped, renamed and relaunched
It revamped, renamed and relaunched what was previously the ParaSet C and C++ group application development environment nearly two and a half years ago now, in June 1995 (CI No 2,676); this after it had told us (March 1995) it had by then secured two rounds of venture capital, totaling $8m, and was saying it would go public as soon as was in profit. It had 45 staff then, 65 by August of 1996 (CI No 2,987), when it had 10 sales offices, 200 customers, and was presumably still in the red. There was also talk of web and Java moves, but these are still not achieved, though it does have an embedded SQL for Oracle string to its bow as well as just C and C++. It still has not gone public, and is keeping mum about about how much VC money it has burned since, but seems to have gone back into growth mode after a quiet year. For example, while it still claims the same number of customers (according to Steven Engel, VP Marketing), it is now saying it has been profitable for the last three years (so why no IPO?), has been growing at the rate of 50% a year, has increased staff from 45 three months ago (dipping from that 65 last year, presumably) to 90, with nine sales offices (down one), and seen the sales force increasing from three to 40 in the same time frame. This coincides with the removal upstairs to chairman of Geisberg in favor of new management talent in the shape of Don Henrich, who formerly occupied a number of roles at Parametric Technology Corp, lastly as senior vice president of marketing, and who became CEO in June.
Ultimate exit strategy
Henrich has led a revitalization of the company, with aims of raising it to a $20m operation next year, with IPO once again being bruited as the ultimate exit strategy. But it strikes one that it won’t get there until it identifies a market sector that is truly hot and growing, and as we all know software engineering in and of itself is just not that. The firm has most recently been touting itself as a year 2000 player, but with the twist of addressing not the Cobol dimension of the problem as the C and C++ end. Engel says the company, though, still sees its mission as being to help attack the software crisis beyond the Year 2000, with the aim not only of the automation of the entire software development process, but the achievement of such longed for CIO nirvanas as improved developer productivity, retention of a company’s intellectual property in its software (i.e., making sure the shape of the code is in some other place than just the departing developer’s head), and generally getting closer to the endpoint of true predictability and repeatability of the whole development cycle. Discover is a family of five application sets, for development, reengineering, configuration management, administration, and documentation, with the aforementioned ancillary Year 2000 tool, Discover Y2K, whose first version was released in September 1996. This in effect creates a persistent databases of applications throughout the enterprise, offering functionality beyond mere repository check-in check-out file facilities, down to individual entity level. Engel contrasts this with something like CASE, in that it integrates with a company’s existing development process (instead of replacing it), and also delivers additional information about what you have and what’s going on. The company, as of July, is now also able to offer Discover and Discover Y2K on NT as well as Unix, with the next general milestone being the imminent release of Discover version 5.1. It points to signs of success such as a recent $1m order for the (non 2000) version of Discover at a manufacturing company and the June announcement by IBM’s Global Services arm that it has licensed Discover Y2K as its Year 2000 solution for C and C++ as objective milestones of the fact that customers are really beginning to accept the need for software engineering.
Preaching to the converted
But is Software Emancipation still just preaching to the converted? Is there really a demand for its technical solutions outside of the kind of team that has probably long since realized the need for something like its product? It identified core markets – telecoms, networking, manufacturing, financial services and embedded systems – which are, after all areas where the idea of software engineering is probably more firmly held than the vast mass of DP out there in end user organizations. However, Engel is adamant that the company keeps talking to managers who have finally seen the light. One suspects that he is right, but perhaps in not enough numbers to really get a Wall St investment banker’s saliva flowing just yet. What will probably happen to Software Emancipation is the same issue facing other reengineering companies, like leading Year 2000 supplier Viasoft Inc. At the end of the day, doing software development right is just not a sexy enough idea for a red herring. It’s taken the enormity of the Year 2000 crisis to finally make us face the need to clean up the data garage – otherwise we’d have all toddled along quite happily enough, naughtily omitting to properly floss, er, document. Software Emancipation may experience greater interest in its intriguing products if it drops the software automation chimera for the time being, makes as much hay as it can while the Year 2000 shines, and use that as the Normandy beachhead to convince managers post 2000 that doing it right is cheaper in the long run. After all, we’re still no nearer real software engineering than we are slimmer’s beer – and believe me, if there were such a product, I’d buy me a lake of the stuff.