I don't have a PDF for these, but maybe you can figure out a way to print-to-PDF from the browser? Here's all the source code for the slides: github.com/oconnor663/bao_presentation
Normally you don't implement a cryptographic function on a language with GC to prevent timing attacks in case GC decided to play in the middle of a cryptographic function.
That's cool for non adversarial situations but should never be used when security is a concern. There's no way this has the same security level as SHA3-Keccak and since it hasn't gone through proper auditing, it's extremely likely to be broken. And for that matter if someone doesn't care about security they may as well simply xor all the file chunks (which is extremely fast. It's also massively parallel because xor is commutative)
Bao is not yet stable, and shouldn't be used in production for that reason, whether or not you believe it's secure. Please don't advise people to use XOR as a hash in any situation. If a bit flips in the same position in two different blocks, the output will be unchanged. That makes XOR a bad hash function even for simple cases like random error detection.
Update: As of January 9, Bao is based on BLAKE3, and its output is stable. BLAKE3 is hot off the presses, though, so production users will probably want to give it some time for public scrutiny. Bao itself has also not yet been audited.
It's not really fair to compare Bao for something it is not. Bao supports verified streaming, for example. SHA3-Keccak does not. You can only hash and verify a whole file. And no, hashing chunks of a file with SHA3-Keccak instead of the entire file is not safe. Bao is not that secure as a whole-file hashing as it supported more operations to begin with.
BLAKE3 is a game changer. I often wonder why it has yet to gain the kind of attention and wider implementation that it deserves?
Hello sir, I need these lectures in PDF format
I don't have a PDF for these, but maybe you can figure out a way to print-to-PDF from the browser? Here's all the source code for the slides: github.com/oconnor663/bao_presentation
@@oconnor663 Thank you so much
i need blake3 implementation in c# .. can you help me .. i am a student
Have you taken a look at github.com/xoofx/Blake3.NET ?
Normally you don't implement a cryptographic function on a language with GC to prevent timing attacks in case GC decided to play in the middle of a cryptographic function.
That's cool for non adversarial situations but should never be used when security is a concern. There's no way this has the same security level as SHA3-Keccak and since it hasn't gone through proper auditing, it's extremely likely to be broken.
And for that matter if someone doesn't care about security they may as well simply xor all the file chunks (which is extremely fast. It's also massively parallel because xor is commutative)
Bao is not yet stable, and shouldn't be used in production for that reason, whether or not you believe it's secure.
Please don't advise people to use XOR as a hash in any situation. If a bit flips in the same position in two different blocks, the output will be unchanged. That makes XOR a bad hash function even for simple cases like random error detection.
Update: As of January 9, Bao is based on BLAKE3, and its output is stable. BLAKE3 is hot off the presses, though, so production users will probably want to give it some time for public scrutiny. Bao itself has also not yet been audited.
For those here, for programmers who do not care about security for their application check out the video "non-cryptographic hash functions"
It's not really fair to compare Bao for something it is not. Bao supports verified streaming, for example. SHA3-Keccak does not. You can only hash and verify a whole file. And no, hashing chunks of a file with SHA3-Keccak instead of the entire file is not safe. Bao is not that secure as a whole-file hashing as it supported more operations to begin with.
@@xwtek3505 why is hashing each chunk less secure that whole file hashing?
I'd rather use Nim