// So we can implement the SAMY hash, which is currently an XOR of SHA-256 with WHRLPOOL
//
// Originally, it was SHA512 and WHRLPOOL, which both have a 512-bit output-size.
// I was then going to cut the result in half and XOR together again. But then I
// though, for now, instead of doing all that extra work, I'll just change the
// two "HashAlgorithms" from SHA512 and WHRLPOOL to SHA256 and WHIRLPOOL.
//
// This was very much easier, as I only had to change the little "512" to say
// "256" instead, and basically the job was done. Of course, this means that OT
// is generating a 256-bit hash in THIS object, and a 512-bit WHIRLPOOL hash in
// the other object. i.e. There is still one 512-bit hash that you are forced to
// calculate, even if you throw away half of it after the calculation is done.
//
// Since the main object has a 256-bit hash, the XOR() function below was already
// coded to XOR the minimum length based on the smallest of the two objects.
// Therefore, it will XOR 256 bits of the WHRLPOOL output into the 256 bits of
// the main output (SHA256) and stop there: we now have a 256 bit ID.
//
// The purpose here is to reduce the ID size so that it will work on Windows with
// the filenames. The current 512bit output is 64 bytes, or 128 characters when
// exported to a hex string (in all the OT contracts for example, over and over
// again.)
//
// The new size will be 256bit, which is 32 bytes of binary. In hex string that
// would be 64 characters. But we're also converting from Hex to Base62, which
// means we'll get it down to around 43 characters.
//
// This means our IDs are going to go from this:
//
//
// To this:
//
//
// You might ask: is 256 bits big enough of a hash size to be safe from collisions?
// Practically speaking, I believe so. The IDs are used for accounts, servers, asset types,
// and users. How many different asset types do you expect there will be, where changing the
// contract to anything still intelligible would result in a still-valid signature? To find
// a collision in a contract, where the signature would still work, would you expect the necessary
// changed plaintext to be something that would still make sense in the contract? Would such
// a random piece of data turn out to form proper XML?
//
// 256bits is enough to store the number of atoms in the Universe. If we ever need a bigger
// hashsize, just go change the HashAlgorithm1 back to "SHA512" instead of "SHA256", and you'll
// instantly have a doubled hash output size :-)
//
bool OTIdentifier::XOR(const OTIdentifier & theInput)
{
// Go with the smallest of the two
const int64_t lSize = (GetSize() > theInput.GetSize() ? theInput.GetSize() : GetSize());
for (int32_t i = 0; i < lSize; i++)
{
// When converting to BigInteger internally, this will be a bit more efficient.
((char*)GetPointer())[i] ^= ((char*)theInput.GetPointer())[i]; // todo cast
}
return true;
}
