[AR] Re: failure recovery (was Re: Re: Flight Controller Features)
- From: Ben Brockert <wikkit@xxxxxxxxx>
- To: "arocket@xxxxxxxxxxxxx" <arocket@xxxxxxxxxxxxx>
- Date: Mon, 18 Jan 2016 12:57:23 -0500
On Mon, Jan 18, 2016 at 12:33 PM, Norman Yarvin <yarvin@xxxxxxxxxxxx> wrote:
The one place where we're pretty much in the same boat as the Apollo
designers is with the software we ourselves write; that'll be as buggy
as ever. Well, even there, there has been _some_ progress in the art
of programming since the 1960s; it's just that I'd hesitate to claim
that modern programming techniques yield more than one order of
magnitude improvement in reliability
The static code analysis software that Carmack has occasionally
pointed at is quite impressive. It's not going to catch a flipped sign
in some obscure corner of a filter in the flight control software, but
it will find a ton of subtle bugs that human eyes gloss over.
http://blog.klocwork.com/static-analysis/analysis-of-unreal-engine-4/
is the most recent.
Rockets are also flying with code generated by computers, which is to
say that the software is written in a deterministic manner by the
shared knowledge of multiple programmers, already debugged on Other
People's Projects.
Ben
Other related posts:
