3 ms·

Playing with symmetric encryption algorithms in Ruby

22 points by damir 6y ago

bawolff 6y ago
> Eventhough AES128 counts as secure today computers are getting more and more powerful. To ensure that our code will stay secure for a longer period we will use AES256 for our following examples. Debatable. AES-256 is more a hedge against the most optimistic views of quantum computers. Its not like a classical computer is going to get fast enough to actually need that. > The different modes of encryption algorithms is out of the scope of this blog post so we will not take a deeper look at them Danger! Danger! The mode is one of the most important parameters. If you do this wrong it will not be secure. > [a bunch of examples using CBC mode without authentication] And this is why discussing modes is important. The examples are insecure because they are using CBC mode incorrectly.
- mothershipper 6y ago
  Should also be using a KDF or PBKDF to generate the key from a password, instead of using the text of the password as the key.
- PeterWhittaker 6y ago
  Great points. Crypto articles written by non-experts always make me shudder. To baldly assert that AES 256 is better than AES 128 without discussing related key attacks is to miss a critical point, as is simply asserting "CBC good" without considering how AES is used; cf expert guidance to use AES CTR in SSH instead, given plaintext-recovery-based attacks. Note: I'm not an expert, just a better-than-average-informed worker near the field. I've had the joy and privilege of working with two expert cryptographers and an expert cryptanalyst, and it was a humbling experience: Brains the size of planets and three of nicest, most decent people you'd ever meet. Side-note: I was working late on a design doc based in part on one of the first PKIX drafts when I came upon something that struck me as odd. The cryptographer responsible for the draft was also working late, so I wandered over to his office and asked about it. Long story short, after we'd drawn and erased a few diagrams on/from his whiteboard, this affable, sweet man fixed me with what I can only describe as a "Cylon stare" and said "So you're saying the lifetime should apply to the key, not the certificate?" When I said "yes", rather more meekly than I'd meant to, the Cylon disappeared and he said offhandedly "OK, I'll update the draft!" I love working with really smart people. Working with smart nice people is sublime.
tashbarg 6y ago
This desperately needs a disclaimer: unless you are only experimenting/learning, keep away from this code. If you're looking for something that you can use in a real project, use SimpleBox from the rbnacl gem.

22 points by damir 6y ago

> Eventhough AES128 counts as secure today computers are getting more and more powerful. To ensure that our code will stay secure for a longer period we will use AES256 for our following examples. Debatable. AES-256 is more a hedge against the most optimistic views of quantum computers. Its not like a classical computer is going to get fast enough to actually need that. > The different modes of encryption algorithms is out of the scope of this blog post so we will not take a deeper look at them Danger! Danger! The mode is one of the most important parameters. If you do this wrong it will not be secure. > [a bunch of examples using CBC mode without authentication] And this is why discussing modes is important. The examples are insecure because they are using CBC mode incorrectly.

mothershipper 6y ago
Should also be using a KDF or PBKDF to generate the key from a password, instead of using the text of the password as the key.
PeterWhittaker 6y ago
Great points. Crypto articles written by non-experts always make me shudder. To baldly assert that AES 256 is better than AES 128 without discussing related key attacks is to miss a critical point, as is simply asserting "CBC good" without considering how AES is used; cf expert guidance to use AES CTR in SSH instead, given plaintext-recovery-based attacks. Note: I'm not an expert, just a better-than-average-informed worker near the field. I've had the joy and privilege of working with two expert cryptographers and an expert cryptanalyst, and it was a humbling experience: Brains the size of planets and three of nicest, most decent people you'd ever meet. Side-note: I was working late on a design doc based in part on one of the first PKIX drafts when I came upon something that struck me as odd. The cryptographer responsible for the draft was also working late, so I wandered over to his office and asked about it. Long story short, after we'd drawn and erased a few diagrams on/from his whiteboard, this affable, sweet man fixed me with what I can only describe as a "Cylon stare" and said "So you're saying the lifetime should apply to the key, not the certificate?" When I said "yes", rather more meekly than I'd meant to, the Cylon disappeared and he said offhandedly "OK, I'll update the draft!" I love working with really smart people. Working with smart nice people is sublime.

tashbarg 6y ago

This desperately needs a disclaimer: unless you are only experimenting/learning, keep away from this code. If you're looking for something that you can use in a real project, use SimpleBox from the rbnacl gem.