@lain it's fun to think about space probe code because it kind of shows that you can break almost every modern rule for good software development and make an exceptional product if you hire the very best people, and highly motivate them through national chauvinism.
@Moon i think people overestimate the complexity of space probe stuff because it's "rocket science", but once you're in space there's practically nothing there and everything runs super basic inertial equations.
@Moon i'm not denying that they wrote a lot of software that worked well, i'm just not buying that they were somehow amazing masters of their craft while we idiot webdevs have to stand in awe before 1960s programmers. there are plenty of cases where software bugs destroyed rockets and probes, both in the US and in the USSR.
@lain survivor bias is real but I believe that before the commoditization of software engineering, engineers were on average of much higher quality. these people still exist in the world, they are just applying their high level of skill to areas that society actually rewards.
@Moon@lain it might also be a byproduct of how directly coupled software and hardware were at that time. Less abstraction meant engineering software required the level of consideration for timings that a mechanic would have working on an engine, where everything has to integrate cleanly into each cycle rather than the luxury of loose timings and scheduling. Thus more of them were Mel as a matter of necessity.
@shibao@Moon he's saying a lot of good things about how to improve the general quality of dev (tests, pair programming, functional programming. i taught CS that way at uni with scheme and it was incredible how much better the code was after just one semester when compared to java or c courses).
But in the end he says that we 'need' to self regulate and become a 'true profession' like doctors and lawyers, which are two of the best examples of what not to become.
@lain@Moon yes i don't agree with him either there, but the history of where computer scientists are pulled from was really interesting to me, i do think we will need some sort of standard like the equivalent of 0 based budgeting like TDD or something to point to to say "we did our due diligence", kind of interesting to me about how in military stuff things have been moving away from "xyz versions work together, all of these versions are Certified" to "we audited how your team works and if these processes are followed then anything you do is Certified", this might be what the whole devops thing was trying to say (except it didn't say anything at all almost)
Add comment