Question

What is the best practice to check that my binary data at the file is not corrupted?

There is some string (lenght + chars) and some MAPs(first size of map, than string).

And when data is corrupted, I get in to an infinite loop, because size of map can be very big value.

Any ideas how to be check and be sure that binary data is not corrupted before/while reading?

Answer 1

The simplest solution is some sort of hash/checksum. There are several options:

• One/two’s complement addition
  • Pros: Very easy to implement – just add up all the 32-bit DWORDs in your data block.
  • Cons: Not very good at detecting bad data – some common bitskews can break it.
• CRC32
  • Pros: It’s fast, relatively easy to implement, has a small hash output size.
  • Cons: Small hash size means there’s a small probability of missing bad data.
• MD5
  • Pros: For most intents and purposes (i.e. not crypto) it’s collision resistant.
  • Much slower than the previous two, and more complex. There are libs though.
• Parity bits
  • Pros: Can be used to detect and correct errors.
  • Cons: Can be difficult to implement properly, increases overall size of data.

Take your pick 🙂