Title: Thinking Aloud

Those of you with long memories might recall that in my last column I discussed the myth of the silver techno-security bullet. Interestingly enough, proof of the example I used, that face recognition technology simply doesn't work, was confirmed dramatically only recently.

The American Civil Liberties Union (ACLU) recently obtained system records of the video surveillance cameras on the streets of Tampa, Florida, using the state's open-records law. The system was deployed in June 2001 and, it turns out, failed to identify even a single individual in the department's database of photographs. In fact, the system is so useless that the police admit they haven't used it since August...

Rarely do my unsubstantiated assertions get so swiftly validated - normally I have to wait years to be able to say 'I told you so!' If you are trying to hold back the tide on this sort of issue at work the URL to download the report from is:

www.aclu.org/issues/privacy/drawing_blank.pdf.

I was going to spend most of this issue on an open letter to Borland about their new licensing conditions. Unfortunately, I fell foul of the cold/cough/chest/flue bug currently doing the rounds, and all intellectual activity ceased while I took to my bed. By the time I staggered back to my computer, Borland CEO Dale Fuller had recanted and issued a mea culpa letter of the sort that normally precedes a session of falling on swords.

For those who missed the whole thing, the more eagle eyed amongst developers spotted that Borland's new license for Kylix included provisions allowing Borland to walk into your house and rummage through your computers and records at their pleasure. It also required you to waive your right to jury trial although it didn't specify what heinous crimes you might have committed that would require trial by jury. About the only thing it didn't include was the right to put your first born to the sword - strange, because I can think of a number of parents of teenagers who would find such a clause very attractive...

Needless to say there was a furore, with many US citizens posting about their rights under 'The Constitution', and others pointing out that 'The Constitution' only covered rights vis-a-vis 'the government', not private companies. (I once had a player, who had been misbehaving, tell me, when I threatened to kick him out, that he had a right to play my game under 'The Constitution'.)

Eventually, Dale Fuller put out a letter explaining that provisions which only applied to corporate volume licenses had accidentally got left in the individual license. Probably this will go some way towards mollifying the developers, but the whole affair has undoubtedly done harm to Borland's image in the developer community.

The whole thing is all the more depressing because Borland were the first people to produce a sensible license for using their software. This was their 'No Nonsense License'. What it said was that you should treat their programs like a book. You can only read a book in one place at a time, and similarly you can only use their program in one place at a time. You can legally put it on as many computers as you like, as long as it is only used in one place at a time. This was the first recognition by any software company that you might want to use the same piece of software on your work and your home computers.

Sadly, I don't think the latest faux pas isolated, it's yet another manifestation of the extent to which Borland has lost its way since the early 90's. I have yet to be convinced that its C++ version of Kylix is anything other than vapourware. I've never felt that about a Borland product before and I first started using their programs with Turbo Pascal, loyally buying Turbo C++ and it's upgrades through Borland C++ to C++ Builder. Indeed, C++ Builder is still my primary Windows programming tool, and I consider that it gives me an order of magnitude productivity boost over using VC++.

Lately, though, I have started to look at alternatives, especially TrollTech's QT, which looks especially promising. Time is running out for Borland as far as I am concerned, and I suspect that I'm not the only one.

On an entirely different tack, I'd like to draw people's attention to an excellent on-line article by Kevlin Henney about exceptions and exception-safe code. I confess to being ambivalent about exceptions. I have the feeling they are used in far too many circumstances that are not, in fact, exceptional... I have suspected for some time that this, and the waste of time in adding exceptions when calling 'new' in Linux, have lead me to neglect this important topic. Reading Kevlin's article was a revelation about what exceptions really are, how they should be used, and what exception safety is really all about.

The article is on the C/C++ Users Journal in their C++ Experts section. Point your browser at:

www.cuj.com/experts/2002/henney.htm to look at the article.

Well just a short column this issue - normal service will be resumed (I hope) next time. Have fun programming.