Rupert Goodwins' Diary

Tuesday 22/2/2005

Trying to keep track of the slurry of outrage, double-dealing and antagonism over software patents is turning into a full time job. So far this week, there's been more confusion in Europe and a shock patent attack on stock market systems -- and it's only Tuesday. Compared to this high drama, the quiet and ongoing unhappiness over just one of Microsoft's dafter applications seems almost invisible. But there's no surprise in the offending idea -- a new BASIC operator called IsNot that says whether two variables live at the same address, roughly the equivalent of "ring up two of your friends and ask whether they're living together". What's amazing is that people still care about BASIC.

In particular, Geoff Perlman (arf!) cares. He's the CEO of Real Software, a company that makes a cross-platform BASIC programming system called Realbasic. He's behind this story, making inflammatory comments about the patent being obvious, having enormous amounts of prior art and failing just about every other test of allowability. Of course, if granted, the IsNot patent would prevent anyone writing a BASIC that would be compatible with any that Microsoft produces - leastwise, not unless they sign up for the standard 'anyone but open source' licence. It must be something to do with promoting innovation, I guess.

But who are these BASIC madmen? Once upon a time, the language made sense -- the B is for Beginners, and the computers of way back when were simple enough that you could get useful stuff cooked up in a few lines of readily understandable code. These days, the sheer amount of complexity you need to shovel into your head before fun happens makes any programming language a large and hairy beast, so why not concentrate on Java or C++, or the more pragmatic grunge of Perl?

Yet there may be a glimmer of sense in the distance. All these new and exciting multicore chips depend for speed on running very clean code out of local cache, and there's never a lot of that. So really efficient performance software will have to be written to the model of small, neat, well-designed modules optimised for speed and low memory usage. In other words, just as it was in the beginning, before enormous applications and flabby operating systems turned your average PC into Lardarse Lounge.

I predict a rapid revival of the old school. With BASIC as our fast prototyping tool and a solid knowledge of machine code, the next revolution will not be GUIized. No wonder Microsoft wants to patent the bleeding obvious.