bad firmware design stuffing up your product

[haxby]

I've just spent 2 days mucking with a wireless router trying various versions of firmware to get it going. Which made me wonder how the one piece of hardware can have over 15 versions of firmware to choose from. If they didn't get it right after the 5th version, who on earth would expect that they would get it right on the next... 16th version?

I wonder if Dave has designed a great hardware based product only to have it stuffed up by poor firmware written by software hacks.

I see a trend amongst PC peripherals where the hardware is well spec'd but the firmware and or drivers are rubbish rendering the hardware useless.

Dave, in your experience, why do you think this is the case? (if you agree with me that is)

You would think it would be more difficult to get the hardware right, rather than the software, wouldn't you?

[charliex]

Well theres different reasons for it, software and hardware bugs, timeframes, standards added or changes. for products like wireless routers especially, if there's some hacked think like WEP they'll often update it.

its a good thing usually, but places will often chuck it out earlier thinking they'll patch it after launch, but this is usually a direct response to the demands of the customers who want it fast and cheap.

[A-sic Enginerd]

Well theres different reasons for it, software and hardware bugs, timeframes, standards added or changes. for products like wireless routers especially, if there's some hacked think like WEP they'll often update it.

its a good thing usually, but places will often chuck it out earlier thinking they'll patch it after launch, but this is usually a direct response to the demands of the customers who want it fast and cheap.

Exactly.

And when it comes to products that use a "standard" protocol there's one more twist to put on it. The explanation I like best is one I heard from a very senior engineer at my old job: it's what it means to be a cable whore.

What he meant by that is, even if your product does everything exactly as it's supposed to per the spec, not every other third party product your products connects with will. And that means sometimes you actually have to do things outside what's required by spec to make it work. If your company is the gorilla in the particular market, and your product gets out early enough, you can actually often become the defacto standard, even if it's not 100% in compliance with the spec. The company I'm at now is classic example of this. They put out a chip for a SATA connection and it was dead on exactly what the spec says. However, almost everybody else product we were trying to connect to had it wrong. Our customers agreed the other companies had it wrong. Even plug fests showed the other companies had it wrong. However, in the interest of expediting our customers products (we just made the chips and firmware, not complete end product), we made the necessary changes to get our product to work with the other POSes.

The more off the wall brand and cheap POS product you go with, the more likely you'll run into this sort of thing. Wireless hasn't quite reached that commodity state (IMHO) like say a 10/100 MAC has so expect to see issues.