Perhaps I should explain why the halting problem is an issue.
In comp sci, the halting problem says we can never guarantee that, given some input, a program won't halt. "Halt" is another way of saying "stop without doing the intended thing," which is what we call a bug.
One of the ways this translates to everyday debugging is that humans cannot ever know the range of all possible inputs or conditions to a program. We don't have that ability any more than we can give someone a list of all the words that can be made with the letters A-Z.
If we cannot write down what all the possible inputs might be, we cannot be sure that one of them doesn't cause halting!
By the same token, as useful as they are, no set of regression tests can prevent all bugs for the same reason: it simply isn't possible to come up with a set of regression tests that is in any real sense "complete" (ie: ensures no halting).
The halting problem was proved mathematically by Alan Turing and applies to all Turing computers, so we know it applies to smart contracts. In fact, the inability of the developers to conceive of one possible input (a zero value from the oracle) is what led to halting in the case of IRON.
To make the problem worse, even knowing that halting (aka bugs) was a possible outcome (a likely outcome, even), not only did the company apparently not seek any outside code audit, they locked down the contract (because it's "law," lol) so that the code can't be fixed even though it's now known to be broken.
So there's the stupidest kind of programming ever. Smart contract is a name in the same vein as the Ministry of Peace in Orwell's 1984. It is anything but smart. It is known to have bugs (halting problem) and they cannot be fixed (locked down "by law," rotflmao).
In comp sci, the halting problem says we can never guarantee that, given some input, a program won't halt. "Halt" is another way of saying "stop without doing the intended thing," which is what we call a bug.
One of the ways this translates to everyday debugging is that humans cannot ever know the range of all possible inputs or conditions to a program. We don't have that ability any more than we can give someone a list of all the words that can be made with the letters A-Z.
If we cannot write down what all the possible inputs might be, we cannot be sure that one of them doesn't cause halting!
By the same token, as useful as they are, no set of regression tests can prevent all bugs for the same reason: it simply isn't possible to come up with a set of regression tests that is in any real sense "complete" (ie: ensures no halting).
The halting problem was proved mathematically by Alan Turing and applies to all Turing computers, so we know it applies to smart contracts. In fact, the inability of the developers to conceive of one possible input (a zero value from the oracle) is what led to halting in the case of IRON.
To make the problem worse, even knowing that halting (aka bugs) was a possible outcome (a likely outcome, even), not only did the company apparently not seek any outside code audit, they locked down the contract (because it's "law," lol) so that the code can't be fixed even though it's now known to be broken.
So there's the stupidest kind of programming ever. Smart contract is a name in the same vein as the Ministry of Peace in Orwell's 1984. It is anything but smart. It is known to have bugs (halting problem) and they cannot be fixed (locked down "by law," rotflmao).