That's why they tend to use timers as well as sensors.
But when red-light were first introduced (and the film was itself very expensive) I seem to remember a load of Melbourne Uni students found one such intersection and spent a quiet evening all standing in the middle "mooning" the camera, while a few of their mates swung stainless buckets filled with nuts and bolts etc low over the sensors, so providing the police with a huge number of very expensive photos to process...

Few signals work based purely on the road sensor, most use timing cycles with input from the road sensors (eg they might adapt their cycle length dynamically depending on how whether cars are queued up at a red and the "green" direction seems to have no cars passing through, or only include a "green arrow" filter when there are actually cars queued up for it) - in fact one of my comp sci projects at Melb Uni was to design a system to do precisely this.

In the late 80's, Melbourne traffic planners tried 2 different high profile schemes (on Maroondah Highway and Canterbury Rd I think) - the first was smart and dynamic sync'ing of the lights depending on actual traffic flows so they'd try to identify a dominant block of traffic (eg rush hour in or out of the City) moving at a certain speed and ensure they caught as many consecutive green lights as possible, adjusting the duration of individual cycles to suit. The other scheme (on a largely parallel road) instead had the lights on a linked but less dynamic timing arrangement (eg there were still "rush hour out of the city" patterns etc but the cycles weren't individually adjusted), but in this case large illuminated signs over the road told you the speed to drive at to catch a green light at the next junction.

Of course both systems have their failings, and I take it the first is now the more common solution, but the latter was an interesting experiment (even if lots of people ignored it, on the basis of the joke about "I don't have to run faster than a lion, I just have to run faster than you").

If you're interested I can drop you an invitation code to

    http://www.mysparebrain.com/

(hit the button to request a pre-launch invite and then ping me on twitter or similar to identify yourself)

It's free and runs in the browser without plugins (but you can export your data locally, an offline mode is fairly straightforward, just not a high priority at this point) - it's aimed more at when you move your files onto the web for storage, but it can handle local files within browser sandbox limitations.

It mixes note-taking, bookmarking, annotation and tagging, reminders, and notifications, but then also features smart integration with other sources, so it knows about things like wikipedia and twitter and actively integrates with such sources via plugins (I'm toying with the idea of a Mendeley plugin).

The aim is not so much to replace lots of other services, but to sit on top of them and give you a smart integration layer, so rather than building a static "bucket" as a repository, you can build something that is more active, and lets you get on with what you need to do with your resources rather than constantly going back and checking them.

And I'm UK based so you may find some of the spellings familiar ("organise"!!)

This generation - don't know they're born etc etc :)

If you can't spot ASCII text strings in a hex dump then you're a tourist - nothing wrong with that, but don't believe that the natives don't understand the local lingo just because it sounds like gibberish to you.

No, the question is not POSIX compliance, the question is the unix philosophy.

Any idiot can emulate some system calls (as Windows Posix shell proudly does), the real question is whether you understood the real point of unix

  - is everything a file, for elegance, for ease-of-use, for consistanecy, for lack of special case hacks ?
  - do you have a small set of tools that do one thing, and apply to everything (because everything's a file) ?
  - can you easily compose such small tools because, at a base level, the way you lauch a new process is a fork style mechanism (whereby a new process inherits file handles etc) ?
  - are all the API's humane (to nick Jeff Raskin's term) - can system calls be interrupted or are system calls considered strangely privileged and all non-kernel code are undeserving scum ?

Windows is so far from being a unix it's laughable to anybody except PHB's (and "developers" who've never worked for PHB's) - and implementing a few stupid POSIX APIs isn't going to make any difference to that.

Whereas the entire "BSD vs Unix" thing I've always thought was pretty fake, BSD is Unix, Linux is Unix, SunOs and Solaris are Unix, I'd pick any of them over a POS like the DOS/Win16/Win32 brain-dead systems I've had to endure for 20 years...