Slashdot Mirror
SpaceX Launch To International Space Station Delayed For Code Tweaks
RogerRoast writes "The first private spaceship launch to the International Space Station has been delayed, possibly by at least a week, the vehicle's makers announced Monday. The commercial spaceflight company SpaceX was set to launch its Dragon capsule atop a Falcon 9 rocket April 30 from Cape Canaveral Air Force Station in Florida." The article quotes SpaceX lead Elon Musk's twittered explanation: "Am pushing launch back approx a week to do more testing on Dragon docking code. New date pending coordination with @NASA."
When is any project ever on time? It's not like they can release beta grade software and release an automatic update to fix it later. If they mess this up, it's going to cost them and maybe, just maybe the engineers plea for proper testing has been answered (a little late though)
"Last minute code tweaks" never go well.
When Fascism comes to America, it will call itself Anti-Fascism, and tell you to give up your guns.
Better they found it now and missed the deadline than went anyhow and exploded. You do not get too many second chances in space.
Space travel has a long history of "extremely rare if not practically impossible" issues coming up to bite you. Missions have been lost because of a single missing comma in the code. So, there is reason for this caution, and neither you nor Elon Musk is going to be able to change it.
Not that SpaceX is infallible - but I think it's NASA behind this requiring an insane amount of testing. They employ the Monte Carlo method of testing which basically tests every (or at least a random sample) value of each input variable and the combinations there of. I don't care who you are, but that method of testing is going to result in "issues" coming to the surface. The problem is that the issues will be extremely rare if not practically impossible. And Musk is not in a position to criticize them, since he wants their business for cargo and crew services.
Not saying this kind of testing isn't valuable, but it doesn't lend itself well to schedules.
This is how you test mission critical systems. No, this is how you must test mission critical systems, regardless of schedules. The key adjective here is "mission critical". This ain't a Heroku web deployment just so you know.
Life support systems require more rigorous testing than simple Monte Carlo. They generally require component testing, bounds case testing, and branch path analysis of the code so that every line of code gets hit during testing.
I've worked on two projects that qualified as life support systems; one was an MRI console for a GE Medical Systems MRI machine (back when it was still being called NMR before it was politically corrected to remove the word "Nuclear"), and the second was a blood gas analyzer. Incorrect operation of the code in either of those cases could have resulted in someone dying as a result of a doctor getting misinformation.
The amount of testing and the rigor of the testing involved in both of those projects was unbelievable. Even then, we were required to carry liability insurance out the wazoo on both projects in case we screwed up the code. There's a reason medical equipment is so expensive.
Space systems that can ram into an occupied space station, and which are intended to some day carry humans to orbit qualify as life support, even if they are being sent up with a load of supplies instead of a human crew. Monte Carlo won't cut it any more than it will for a system call fuzzer trying to find a sequence of three system calls in a row that , if they are called with precisely the right parameters, will trigger a kernel panic.
-- Terry
NASA and Russia are extremely cautious when it comes to anything ISS-related for very good reason. If this thing really screwed up and seriously damaged ISS to the point where they had to abandon it, it would probably end the era of human spaceflight and lead to big budget cuts for both agencies.
What political party do you join when you don't like Bible-thumpers *or* hippies?