Dreaming of a faster future

Written by David Tebbutt, MacUser 10/91 item 02 - scanned

Right now, I'm cheesed off with graphical user interfaces. They might be the best thing that ever happened to computers, but you try developing applications for them. You can get a long way into coding before you realise that your brainchild is going to annoy the hell out of its user. And the GUI is entirely to blame. If that hadn't been invented, we could all be writing punchy little programs that go like the clappers, even on the most basic machine.

Thanks to Xerox and Apple, developers now have to devote much of their programming energy to minimising the effect of bitmap displays on code. Either that or they have to persuade users to buy more powerful machines, which is more or less the route that Microsoft has taken with Windows 3.0. The PC on which I run Windows has an 80386 DX processor, 4M RAM and is now waiting for a hard disk upgrade because 40M is no longer enough.

My Mac SE with 4M of RAM and a 40M hard disk copes better than the PC, but I still encounter the odd program that runs like cold treacle. In fact, I have spent a large chunk of the last 24 hours struggling with two of these slow-running programs. One was written by me, the other by a friend.

Mine was written on the PC, using Microsoft's Visual Basic. His was written on the Mac, using the scripting language of Blyth's Omnis database program. Each program looked pretty good until we wanted to vertically scroll our windows. My friend, rather than assume the presence of a particular font, used bitmap text images, which made the screen creep up and down really slowly. The scrolling on the PC wasn't much better, so I assume that it, too, was treating the window as a bitmap. If it wasn't, then Windows is worse than I thought...

Neither program is finished but even if they were, we couldn't inflict them on the world at large because they are so slow. In both cases, we have layer after layer of alien code between our application and the machine, and it's this that soaks up the power. Had we written our programs in assembly language or C, we'd have a much tighter relationship between the program and the machine. Performance would have been improved, but our programs would have taken much longer to write.

This poses something of a dilemma. We'd both like to get our programs on the market, but neither of us can afford to become the laughing stock of our users. Nor are we rich enough to throw a huge team of C programmers at the problem.

Still, our problems aren't really your problems, except that I assume you too would like computers to do what you want, rather than what some software whizz kid thinks you should want. For this to come about, tools need to be developed that will enable you to throw together applications at the drop of a hat. This means even more layers of code between you and the machine. And, if we're having speed problems, yours will be worse.

There is no simple software solution. To make life easy, you have to have a graphical user interface. You need a way of getting your computer to do what you want, in the style you want, at the speed you want, and with the minimum of effort. This means masses of software on tap to translate your sketchy ideas into working programs. And the only way this can possibly come about is to make beefy machines a standard.

This has been happening in the PC world. The 80286 was the standard machine just a couple of years ago. Now, no self-respecting buyer will consider anything less than an 80386. I've just tested my 80386 against an old IBM/XT. My disk drive gives a 10 times better performance, it computes 22 times faster, and overall performance is 18 times greater. Yet I still don't have enough power.

Perhaps this is why everyone's getting excited about RISC. With other specialised processors which relieve the pressure on the CPU, you could be talking about very fast machines indeed. They'll start off being called professional workstations and cost an arm and a leg. But within a few years, they'll be de rigeur in the office. Within five years, there will be commodity products selling for the equivalent of 1000 today. Our frustration with machine speeds will disappear.

But by then, we'll regard scrolling graphics as old hat, and we'll be after video-conferencing, speech recognition, virtual reality or some other power-hungry gimmick. Just as manufacturers deliver what you thought you wanted, at a price you can afford, they'll think of something new to whet your appetite.