iOS 13 notification 'text bomb' crashes iPhone, iPad

2»

Comments

  • Reply 21 of 22
    avon b7avon b7 Posts: 8,327member
    lkrupp said:
    Rayz2016 said:
    This happened once before. They fix it. And it happens again with a different set of characters?
    Someone’s code needs to be pulled out and rewritten from scratch. 
    And the new code would still have bugs. Software engineers and coders don't have the time or inclination to spend hundreds of hours trying  random, different key combinations to find one that exposes a bug. That’s for the wack jobs who sit around all day, every day, in their parent's basements, fiddling around. 
    The problem isn't simply the 'bug'. It is that it can bring the entire system down so easily. We should be beyond this particular type of problem. 
     0Likes 0Dislikes 0Informatives
  • Reply 22 of 22
    razorpitrazorpit Posts: 1,796member
    svanstrom said:
    razorpit said:
    I was thinking, why can’t there be a sandboxed function that every incoming text And paste command gets passed through. If the function returns a “0” the text message is displayed or the paste completes. If it returns a “1” then the text is forwarded to Apple and the message to the user is killed. 
    Practically speaking that's just pushing the problem into another layer, while at the same time saying that that layer can't have any bugs; and then when a bug surfaces in that layer, then you'd say the same thing about another layer that is supposed to not have any bugs in it, and then another, and another… So practically speaking your idea is essentially the same as you simply saying "I was thinking, why can't there be no bugs anywhere".

    Sure, someone will now come to your rescue and claim that what you're describing is some simple form of if…else or try…catch, but that's simply not understanding the complexity of programs beyond the most basic things in the language; because unless written by the most basic amateur there's already all kinds of catching of errors and data validation and data being cleaned and data being sandboxed and whatnot, but… then there are the bugs that couldn't be predicted, including problems with the compiler or language itself, or a programmer mistyping something, or an unfortunate case of autocorrect, or a disgruntled employee, or test code accidentally being used live, or a bad copy-paste or cmd-z or about another gazillion other things that can't be predicted and avoided or caught.

    Edit: Imagine that the bug is actually in how the display in a very unique situation deals with drawing the shading on somewhat overlapping characters; so it might not be possible to trigger at all unless you in the testing actually include a return code/message from the actual physical display, so in your testing it theoretically is possible that you actually need a physical device to trigger something. So to test everything you'd actually need at least one more physical device that first tries to do everything before it's done on your device. Two phones, with yours always pausing to let the other device try everything first. And even then there's always a layer somewhere that can't be tested, and that might contain bugs.
    Good points and well-written.
     0Likes 0Dislikes 0Informatives
Sign In or Register to comment.