[Elephant_Canyon]

Yes. Why should software companies set a goal to design bug-free software and apologize when they don't?

If it's so easy to design bug-free software, why don't you show them how it's done?

[Richard Buckley]

I think the biggest difference between any other product and software is that with any other product people don't go that extra mile to find ever more ingenious ways of breaking it.
I don't think it's the fault of the software engineers that it keeps happening, especially when you have to integrate your product with 3 parties who at times they have no control over.
Imagine if you bought a bicycle and then someone said if you lay it on its side on a wall and add a half ton weight to each side the frame will bend in half then get told to fix it. Your first question would be why would someone do that but because some bed wetter has you have to fix it. The next time they'll use a one ton weight.
Software Devs can't anticipate every move and like you say they can't fix it until it's broken which is already too late.
Get your phone direct from BlackBerry and have more control over the OS was the mantra of BB10. Ironically people would rather have Snapchat so I guess they can't complain now. You pays your money and takes your chance.

A much better example would be that you build a car with a faulty ignition and steering lock that can fail and cause loss of control of the car. You know about it and decide not to fix it until people start dying because of it.

In Android the Multimedia Library - formerly known as StageFright until they changed the name to try to distance it from the publicity - has had serious bugs patched in every monthly update all the way back to the first StageFright proof of concept. As a professional software developer this tells me that the original library was very poorly written and something needs to be done about it. As we now know because of a couple of studies the number of independent discovery of zero days is quite high so we should not just be glad they are fixed because we don't know how many people knew about them before the patches were released, or how many bugs are known that won't be found and patched until months from now.

BlackBerry knows this. Have a look at the first duty of the job posting.

http://blogs.blackberry.com/2016/03/...-free-puppies/

https://www.linkedin.com/jobs/view/267894023/

LeapSTR100-2/10.3.3.2205

[sorinv]

If it's so easy to design bug-free software, why don't you show them how it's done?

Would you like some fries with that?
Cost-free, too?
Do you want me to pay for the disclaimer? I'll do that, too, since you are so nice.

Posted via CB10

[Richard Buckley]

If it's so easy to design bug-free software, why don't you show them how it's done?

Formal verification has been around a long time and is used by hardware manufacturers who face severe losses since it is practically impossible to patch an integrated circuit. Writing bug free software is not impossible, it just isn't as profitable since most people already believe it is impossible and therefore expect bugs.

https://en.m.wikipedia.org/wiki/Formal_verification

LeapSTR100-2/10.3.3.2205

[sorinv]

Formal verification has been around a long time and is used by hardware manufacturers who face severe losses since it is practically impossible to patch an integrated circuit. Writing bug free software is not impossible, it just isn't as profitable since most people already believe it is impossible and therefore expect bugs.

https://en.m.wikipedia.org/wiki/Formal_verification

LeapSTR100-2/10.3.3.2205

This!

I assume, although I am not in the field, that in some critical applications like aerospace, even the software designers do proper verification and do not "tape out" with known flaws.

Posted via CB10

[Richard Buckley]

This!

I assume, although I am not in the field, that in some critical applications like aerospace, even the software designers do proper verification and do not "tape out" with known flaws.

Posted via CB10

And others. I recently heard about an open source library written by a mathematician who does formal proofs before coding.

But for consumer grade software and especially Web 2.0 the methodology of the day is Agile and the philosophy is ship now patch later. Sure there is test driven development, and continuous integration. But that is just writing software to test other software. If your test software has as many bugs as the software it is testing you can't hope to find them all. But the Agile Manifesto can't be wrong, or can it?

I saw an article in which the author estimated software has one bug for every 100 lines of code. That's 10,000 per million lines. Would you want to board an airplane with 70,000 software errors? Although apparently a Fort GT would have 100,000 errors, and an new F150 1.5 million errors. I think he's been looking at really crappy code, but still the mind boggles. When I was in college you wouldn't pass the course at that rate. But then computer time was way more expensive than programmer time, the last thing you wanted to happen was have an error crash your run which meant you got nothing out of it.

It is very rewarding writing bug free code that continues to do its job for years after the last time you looked at it. Very valuable for employers too once you convince them how much better writing high quality code can be for their business. It's a good thing I'll be retiring soon.

LeapSTR100-2/10.3.3.2205

[sorinv]

Yes, the days when you were not allowed to make ANY mistakes, in school and at work, even when writing software, are long gone.
Now, making mistakes is something to be proud of. The more the merrier. In hardware, you get fired, because it is far too expensive to fail.

Posted via CB10