3 ms·

I hate articles like this. They're so one sided. Here's a money quote: "Errors in complex code are inevitable". Even the headline sets the same tone and calls t

I hate articles like this. They're so one sided. Here's a money quote: "Errors in complex code are inevitable". Even the headline sets the same tone and calls the flaw a "fluke". Oops, this programming stuff is hard, be thankful that it "works" at all. Bullshit. There's something that was left unsaid in the article, specifically "best practices". Why wasn't the length validated at all? There's nothing new or "complex" about simple defensive programming. How can anyone (even a part timer) working on software that's so security critical be so clueless? Forget about more obscure stuff like the full disclosure mailing list, just reading CERT alerts should make this abundantly clear to anyone in security. Hell, the xkcd cartoon [1] makes it abundantly clear. If you can't take that cartoon to heart, you have no business writing Internet facing software. I think Marco Peereboom got it right oh so many years ago when he said that OpenSSL was written by monkeys. [2] However, the article does get something right. It's insane that something so critical to internet commerce is essentially a hobby project by a few people mostly in their spare time. That's not simply crazy, that's totally fucking insane. That's the biggest takeaway of this entire fiasco. [1] http://xkcd.com/327/ http://xkcd.com/327/ [2] https://news.ycombinator.com/item?id=7556407 https://news.ycombinator.com/item?id=7556407

stcredzero 13y ago
Bullshit. There's something that was left unsaid in the article, specifically "best practices". If the NTSB used the "best practices" mantra like the software industry did, then there would be no stick shakers, ground proximity warnings, and collision avoidance systems. Thousands more would be dead without such automation. It's amazing that the software industry so often responds to such problems by saying we programmers should do one more thing, and do it consistently without fail. (Then no concrete steps are taken.) Aren't programs better suited to such tasks? We should be using languages or libraries that are memory-safe and give us automatic means of avoiding timing attacks. When we in the software industry fix widespread safety issues, our answer shouldn't be, "we should've done X, those people are to blame for not doing X, let's all just do it from here on out." If we were fire marshals, there would be no mandatory fire alarms, emergency exits, and occupancy regulations. Instead we'd "tsk tsk" and talk about how those poor victims were "doing it wrong." If governments were exploiting this bug, they were doing the opposite of what they should've been doing: fixing it once and for all. I've read several users comment that an Ada implementation of TLS would avoid the same class of errors entirely.
- PhantomGremlin 13y ago
  > Aren't programs better suited to such tasks? You're absolutely right. "Wetware" is highly imperfect; widespread adoption of better tools is essential. > I've read several users comment that an Ada implementation of TLS would avoid the same class of errors entirely. I don't know Ada, but at least according to Wiki it is much more suitable for this sort of programming [1] than using C. So why isn't Ada more popular? Perhaps its time has passed? I do see frequent mention here on HN of other, more recent languages with similar goals. [1] http://en.wikipedia.org/wiki/Ada_%28programming_language%29#Features http://en.wikipedia.org/wiki/Ada_%28programming_language%29#...

stcredzero 13y ago

Bullshit. There's something that was left unsaid in the article, specifically "best practices". If the NTSB used the "best practices" mantra like the software industry did, then there would be no stick shakers, ground proximity warnings, and collision avoidance systems. Thousands more would be dead without such automation. It's amazing that the software industry so often responds to such problems by saying we programmers should do one more thing, and do it consistently without fail. (Then no concrete steps are taken.) Aren't programs better suited to such tasks? We should be using languages or libraries that are memory-safe and give us automatic means of avoiding timing attacks. When we in the software industry fix widespread safety issues, our answer shouldn't be, "we should've done X, those people are to blame for not doing X, let's all just do it from here on out." If we were fire marshals, there would be no mandatory fire alarms, emergency exits, and occupancy regulations. Instead we'd "tsk tsk" and talk about how those poor victims were "doing it wrong." If governments were exploiting this bug, they were doing the opposite of what they should've been doing: fixing it once and for all. I've read several users comment that an Ada implementation of TLS would avoid the same class of errors entirely.

PhantomGremlin 13y ago
> Aren't programs better suited to such tasks? You're absolutely right. "Wetware" is highly imperfect; widespread adoption of better tools is essential. > I've read several users comment that an Ada implementation of TLS would avoid the same class of errors entirely. I don't know Ada, but at least according to Wiki it is much more suitable for this sort of programming [1] than using C. So why isn't Ada more popular? Perhaps its time has passed? I do see frequent mention here on HN of other, more recent languages with similar goals. [1] http://en.wikipedia.org/wiki/Ada_%28programming_language%29#Features http://en.wikipedia.org/wiki/Ada_%28programming_language%29#...