Marine pilots on USS America speak out about flying the F-35B

a reply to: Zaphod58

Let me rephrase that. It might be accurate. But I don't think that's the whole story. There's got to be more to it than that. At least that's what sticks out in my head.

As for using C++. Yeah. I can see that. There's a bunch of debug tools and unit testing tools out there. Now you're making me wonder if they are going out to github and getting code from open source repositories. That would be interesting.

And I looked at ADA. Yikes. I wouldn't want to program in that.

a reply to: grey580

Why does there have to be more to it? I know a lot of people have a hard time believing it, but sometimes the military is incredibly backwards and what you're looking at is exactly what it appears to be.

a reply to: Zaphod58

What an incredible partnership of man and machine! It seems the added tech has actually increased the interaction of the pilot. They're more involved in their environment, rather than more detached.

Instead of supplanting the human element, this tech is augmenting the human element - and vice-versa.

a reply to: grey580

What would that "more" be in your opinion? Do you think there aren´t capable people there so they would pull third party sources? For what reason exactky? Those people hired there are not script kiddies, they know precisely what they do and they have to keep standards that most github projects would certainly not fulfill. They have their own repositories, probably tripple tested and what not, until it get´s green light to be used in one of the most advanced machines mankind ever built.

And if it´s about the report that said there are so many bugs in the software, some time ago. It´s very seldom that you hack a program together, compile it and will never have to change anything. Some bugs turn up over time, some are extremley hard to track down, even with the right tools. We talk about 4? million lines of code, in the end. That´s not something you hack together by pulling 3rd party repositories and probably relying on properitary DLLs you can´t tell what´s in there, if you don´t disassemble it and even then, you have assembler code. There are whole contests about hiding stuff in code you´d never find out, let alone grasp the meaning of it.

That´s the difference when I string something together in private, I don´t develop, I do it on the fly. Those are not programers that get told "make this happen". Software engineers planed the single parts and you get a strict frame what you are supposed to deliver in code. There´s no #ing around like a hobby programer. The printed out version of the programs in readable text + the actual source + documentation and conventions would keep a single printer very busy for some time. This stuff takes up rooms.

Not going to happen. Just saying.

originally posted by: Zaphod58
a reply to: In4ormant

At this point, no. There will eventually be UAVs capable of doing the mission that the F-35 will do, but right now, there isn't. If they tried, they'd lose them all, over and over again for no gain. I don't see why that's so hard to understand. You're sending a signal across thousands of miles, which means there's a delay. Until they figure a way around that, there's no way to have an air to air capable UAV that can fight its way in, and hit the target.

there's only one way to eliminate the delay - AI on board. and as much as i think it's possible today (it doesn't have to be an AI per se - just some piece of code smart enough to evaluate the situation and make proper decissions), machine making the decission to kill is probably the part that won't fly. maybe AI-assisted fly controls (making decissions to avoid critical threats while waiting for delayed input) would be enough, but i imagine that's already on board. if it's not though - that's the most obvious way to go for UAVs. not saying that F35 isn't worth its money, far from it, just throwing in my 5 cents.

originally posted by: Maxatoria
Even C/C++ now is becoming like COBOL as a legacy thing as universities just don't teach it quite often as its not fashionable.

when it comes to programming, universities are useless. tons of new software every year gets done using C/C++. in case of console games for example, you often don't even have a choice, since those are the only languages supported by some of the SDKs.

as for finding people who know any particular language when it goes out of fashion at universities, it may be harder, but at least you're guaranteed to find people with some real knowledge and experience, not some wannabe-programmers that know only what they've been told at university. limiting yourself to people with degree and no real experience, no passion for self-learning and no original projects of their own, is a fastest way to fail.

so whatever knowledge regarding programming languages you have from university, do yourself a favor and throw it out of the window. C isn't going anywhere and until the whole world will switch to something like quantum computers, either requiring different basic concepts than those provided by C or having such abundance of processing power that software using only 10% of it will still be running too fast for any regular user to notice, C will remain in business.

a reply to: Zaphod58

I think my comments long long ago about the reports (negative) about the F-35 were basically hit pieces, like saying an unloaded gun is not as good as a sword, so guns should be abandoned.

a reply to: jedi_hamster

right, microcontrollers rule the world, aint going to be replaced with a fancy new language. C/C++ are one of the most powerfull higher languages.

a reply to: jedi_hamster

Is it going to be you on the ground calling in an airstrike when your AI equipped UAV has a rounding error when dropping it's bomb?

I was a SAR (Search And Rescue) crewman. It was my job to ride in the back of a slow, usually unarmed helicopter, to try to pick up a downed pilot. Our aircraft was hit by ground fire on more than one occasion (Lebanon 1983).

There is nothing that I would like to see more that a system that will let us make a precision strike on highly defended targets without having to risk pilots or aircrew. Nobody getting shot down means not having to risk rescue assets.

The technology just isn't there yet. It is all fine and dandy to sit there and be in favor of the newest whizzbang gadget in military hardware. You might feel a little different if it was your ass on the line needing something made by the lowest bidder to bail you out.

Does anybody remember the Army unit hit by friendly fire in Afghanistan? They were providing coordinates for an airstrike by JDAMS when the battery in their GPS died. They replaced the battery and gave the bomber the coordinates. The problem was that the GPS didn't have a memory backup. Instead of giving the coordinates of the enemy's position, they gave their own.

originally posted by: Zaphod58
a reply to: Maxatoria

I think the F-35 program is going to drive home that they need to change to a COTS system. The problems they've had have been resolved quicker by using C instead of just ADA.

That would be surprising to me. C and C++ have far more holes and gotchas and sources of subtle error vs Ada, which was designed precisely for high-reliability embedded tasks. With the amount of money devoted to the overall project, paying a compiler team to make more tools for Ada would have been a trivial expense.

a reply to: mbkennel

It may not necessarily be C/C++, but some kind of commercial chipset and language.

a reply to: Zaphod58

Always seem to remember a saying that went something like

C Will provide everything you need to hang yourself

C++ Will provide the rope and the rest you have to provide yourself.

But on custom designed hardware i'd imagine theres very few options as the time it would take for verification such as memory leaks etc would take a long time and a lot of effort.

Always remember in the late 80's having to debug semi-compiled code..trying to make sure the compiler aint messed up something subtly is a job for someone with an IV line of caffeine especially once you go over about 30 mins as trying to keep it all in your head aint easy.

A saying CAD Monkeys say,"Error at Keyboard."

a reply to: Blackfinger

As a "CAD monkey", I believe that it is "An error in the chair keyboard interface."

PEBKAC
Problem exists between keyboard and chair

originally posted by: Maxatoria
a reply to: Zaphod58

Always seem to remember a saying that went something like

C Will provide everything you need to hang yourself

C++ Will provide the rope and the rest you have to provide yourself.

But on custom designed hardware i'd imagine theres very few options as the time it would take for verification such as memory leaks etc would take a long time and a lot of effort.

Always remember in the late 80's having to debug semi-compiled code..trying to make sure the compiler aint messed up something subtly is a job for someone with an IV line of caffeine especially once you go over about 30 mins as trying to keep it all in your head aint easy.

proper test suite and something like valgrind should do the job, for memory leaks at least. processing power isn't an issue anymore, so a lot of such tests can be automated today.