It feels like it would be *SO* easy for the cryptography community to define some basic file format spec containing a header with magic bytes, an IV and an HMAC, probably using aes_gcm or something, and then declare that as the standard general-purpose encrypted file format. But nooo, they can't make specifications, they *have to* tell you to use their shitty C implementation in OpenSSL or command line tool
@mort I think they have?
https://github.com/FiloSottile/age
Spec: https://github.com/C2SP/C2SP/blob/main/age.md
Other language implementations:
- Rust (lib & CLI): https://github.com/str4d/rage
- TypeScript: https://github.com/FiloSottile/typage
- Python: https://github.com/woodruffw/pyrage
- Java: https://github.com/exceptionfactory/jagged
- WIP Kotlin: https://github.com/android-password-store/kage
- WIP Swift: https://github.com/jamesog/AgeKit
I do agree more language implementations are needed -- notably, it would be great to have high-quality C-bindings wrapping the Rust age crate ("cage" anyone? such a great project name...). But I think this is mostly a matter of time to fill out the breadth of implementations?
@mort i think key formats are really important and there's an unofficial standard emerging from tools like signify and age
@mort i think encrypted file formats seem really difficult to standardize and i'm curious what use case you're envisioning
@mort i also think compression is done completely wrong and so are checksums