[@sounisi5011/encrypted-archive]: Future updates, including breaking changes

[sounisi5011]

It has been a year and a half since I first released @sounisi5011/encrypted-archive. Due to the long time that has passed, I would like to make an update that includes breaking changes to the archive data binaries.

• Cannot add new fields to archive data that are required for decryption.

  The archive data is the Protocol Buffers. So if we add a new field, the old @sounisi5011/encrypted-archive can parse the new archive data. However, if the new fields are metadata required for decryption, the old @sounisi5011/encrypted-archive will throw the following error:

   Unsupported state or unable to authenticate data
  

  This error is not helpful; @sounisi5011/encrypted-archive MUST throw an error like "This archive is not supported".
  To detect this, the archive data should contain a version number.
  However, there is no such field in the current archive data.

• I want to add additional binary data required for decryption

  In @sounisi5011/encrypted-archive, counters are used to avoid IV reuse.
  However, to absolutely increment these counters requires synchronization between multi-threads, etc.
  A way to reliably guarantee this would never be achievable within the Node.js library.
  (Even if we provided an API for users to synchronize this counter, I doubt that users would use such a complex API correctly).

  By the way, the condition under which IVs should not be reused is when the "key is the same value".
  In other words, by providing additional input (e.g. current time, PID, MAC address, etc.) to the key derivation function, "IV reuse" will not occur, even with the same password and the same IV.

  By adding a field containing such data, @sounisi5011/encrypted-archive would not require a complex implementation to synchronize IV counters.

• I want to remove the compress_algorithm field

  See [@sounisi5011/encrypted-archive] Maybe the compress option should be deprecated #661
  If the compress option can be removed, then the compress_algorithm field is also unnecessary.

• Weak Argon2 option should not be allowed in archive format

  If the format specifies minimum values for the time_iterations and memory_kib fields, we should never be able to use the weak Argon2 option.
  Something similar is already done for the crypto_nonce_counter_add_or_reset field in the SimpleHeader message.
  See [@sounisi5011/encrypted-archive] Disallow weak Argon2 options #434