Hacker News Reader

Comment by xvilka | Hacker News Reader

Why not use the Rust implementation[1] of the PGP instead? I feel that writing something new and complex like this in non-statically typed language (Python in this case) is a recipe for disaster. Someone suggested using TUF, but it's also in Python and prone to shifting compile-time errors to runtime.

[1] https://sequoia-pgp.org/

View on HN

Because even memory-safe implementations of PGP are saddled with the complexity and legacy cruft the the protocol itself, and the thing Debian projects are actually looking to do is exactly what the signify/minisign family of signature schemes does, using exclusively modern cryptography, with the simplest possible interface.

Makes sense if they don't plan it to become more accommodating and complex over time.

Ironically, by using a modern, stripped down signing system like this, they've probably increased their opportunities to do interesting things down the road, since they don't have all the possible weird interactions with the PGP system to consider.

Still have to trust 1k RSA keys because GPG does not allow us to say they're unsafe.

i am pretty sure you can mark them yourself. but that they are generally still used, well that's called backwards compatibility. and that's needed.

You can tell how long a key is just by looking at it. You can then say anything you want. The use of particular keys is a Debian policy issue. It has nothing to do with the use of a particular message format like OpenPGP.

That's not how it works, no. This is a gpg specific issue as it does not communicate the key size over the status fd.

We don't parse the keys ourselves for obvious reasons.

Dunno about that but --with-colons certainly works to show the key length of a key.

i would bet that if debian had some sane requirements, sequoia could fulfill them.

sequoia is a rewrite which avoids the legacy cruft of the old OpenPGP. and they already have a library, a cli tool and tests

That's an absurd statement. OpenPGP _is_ the legacy cruft. It's made for _emails_. It makes no sense elsewhere. The encoding for clearsigned signatures is overly complex.

And Rust is inappropriate in any case for now, it's not allowed in Ubuntu main.

It's a problem of Ubuntu then, not Debian.

Not gonna have two different signing implementation. Same security concerns apply to Debian too, of course.

Having 130+ packages to rebuild for a critical security update is no fun.

> That's an absurd statement. OpenPGP _is_ the legacy cruft. It's made for _emails_. It makes no sense elsewhere. The encoding for clearsigned signatures is overly complex.

No. It's a message encryption, and namespaces are emails, but then, most logins nowadays are identifiers via some email. this is not true. and encodings are fixable and doable in a good library, so there's no misuse.

> And Rust is inappropriate in any case for now, it's not allowed in Ubuntu main.

that's new to me, but possible, why? any link?

Rust is a security nightmare. We'd need to add over 130 packages to main for sequioa, and then we'd need to rebuild them all each time one of them needs a security update.

Does Debian enforce a particular implementation of OpenPGP? For all we know Sequoia is already in use. OpenPGP is valuable mostly as a popular and widely standardized format. There a lots of implementations.

You have the choice between GnuPG 1 and GnuPG 2 :D

Reply on news.ycombinator.com