4 ms·

The F35's avionics suite is not just for controlling the aircraft (which is difficult in itself), it's for controlling, analysing and displaying the data from s

by berkut 12y ago

The F35's avionics suite is not just for controlling the aircraft (which is difficult in itself), it's for controlling, analysing and displaying the data from sensors (radars, RWR, IFF, infra-red, etc) on an integrated display system, data transfer (can transmit its own sensor info to other aircraft and receive it from others), and interfacing with a variety of NATO weapon systems. The F35's integrated system design was a (possibly too much) very audacious design, and includes over 10M lines of C++. It was questionable why C++ (instead of using existing ADA software) was used from the "re-inventing the wheel" perspective, but existing software in ADA very likely wasn't good/sophisticated enough and would have required significant re-writes anyway. Over the past 10-15 years there has also been a huge move away from ADA to more COTS languages/solutions by management in order to "save money".

Someone 12y ago
Software can always be improved, and this may need it, but this headline and my remark are specifically about the gun (also: the plane is flying, isn't it? If so, I assume that all those detectors, displays, etc. already work to sufficient extent) Reading http://www.thedailybeast.com/articles/2014/12/31/new-u-s-stealth-jet-can-t-fire-its-gun-until-2019.html http://www.thedailybeast.com/articles/2014/12/31/new-u-s-ste..., the plane carries so few bullets that it may be necessary to make its shots count, but even that shouldn't be that hard that it takes years and years to develop the software to shoot a gun. Caveat: I know next to nothing about what the U.S.A. expects these guns to do or what they can do. My model of what a gun does is "spew out lead when the pilot presses a button", with very few refinements. A video camera that stops shooting when the original target goes out of range and a friend-or-foe detector input that stops shooting, would be nice additions, but since the missile system works. I would guess that the 'interfacing with a variety of NATO weapon systems' part shouldn't be that hard for the gun part of the plane's software. Or are these guns almost auto-firing and do they want to shoot at enemy vehicles that are in close combat with friendly forces, and only hit enemies? Maybe even have the gun control steering of the airplane (target turns left, please follow, so that I can keep firing)? That would seriously complicate matters. Or is this a matter of bureaucracy in action in that this module missed a shipping deadline and now has to wait for the next software release for the entire plane that will come in 2019?
- berkut 12y ago
  It probably would only take a month to get software to "just" fire the guns working. But the defence industry I believe (left it 7 years ago, but still know people in it) still largely use the waterfall model for designing software a lot up front, and for safety-critical software, there's a lot of static analysis done on the code, so you can't just write it and ship it. I think what probably would take so long is integrating this new gun firing module into the entire software suite, or as you say, waiting for the next software version.
engi_nerd 12y ago
Here is info on the subset of C++ that was mandated for use in the F-35: http://www.stroustrup.com/JSF-AV-rules.pdf http://www.stroustrup.com/JSF-AV-rules.pdf

by berkut 12y ago

Someone 12y ago

Software can always be improved, and this may need it, but this headline and my remark are specifically about the gun (also: the plane is flying, isn't it? If so, I assume that all those detectors, displays, etc. already work to sufficient extent) Reading http://www.thedailybeast.com/articles/2014/12/31/new-u-s-stealth-jet-can-t-fire-its-gun-until-2019.html http://www.thedailybeast.com/articles/2014/12/31/new-u-s-ste..., the plane carries so few bullets that it may be necessary to make its shots count, but even that shouldn't be that hard that it takes years and years to develop the software to shoot a gun. Caveat: I know next to nothing about what the U.S.A. expects these guns to do or what they can do. My model of what a gun does is "spew out lead when the pilot presses a button", with very few refinements. A video camera that stops shooting when the original target goes out of range and a friend-or-foe detector input that stops shooting, would be nice additions, but since the missile system works. I would guess that the 'interfacing with a variety of NATO weapon systems' part shouldn't be that hard for the gun part of the plane's software. Or are these guns almost auto-firing and do they want to shoot at enemy vehicles that are in close combat with friendly forces, and only hit enemies? Maybe even have the gun control steering of the airplane (target turns left, please follow, so that I can keep firing)? That would seriously complicate matters. Or is this a matter of bureaucracy in action in that this module missed a shipping deadline and now has to wait for the next software release for the entire plane that will come in 2019?

berkut 12y ago
It probably would only take a month to get software to "just" fire the guns working. But the defence industry I believe (left it 7 years ago, but still know people in it) still largely use the waterfall model for designing software a lot up front, and for safety-critical software, there's a lot of static analysis done on the code, so you can't just write it and ship it. I think what probably would take so long is integrating this new gun firing module into the entire software suite, or as you say, waiting for the next software version.

engi_nerd 12y ago

Here is info on the subset of C++ that was mandated for use in the F-35: http://www.stroustrup.com/JSF-AV-rules.pdf http://www.stroustrup.com/JSF-AV-rules.pdf