[staticassertion]

> Writing unsafe Rust instead of unsafe C is not giving you any real advantage for the existing projects, as long as the state is as it is.

Your argument hinges pretty heavily on this. It's incorrect. The reality is that unsafe code in rust is not like unsafe code in C. First, it's bounded to a module. Second, you can grep for it (huge). Third, you still have less UB in unsafe rust than in C.

I've never written code in rust that required unsafe, after many thousands of lines.

> Is safely rewriting a crypto library which is useful in real project still a too big task to be taken?

https://github.com/ctz/rustls

?????????????

> But I'd like to see a real life case of its success too.

Ripgrep is a nice one to point to. There's redox too. Tons of rust code out there.

[acqq]

> It's incorrect.

> I've never written code in rust that required unsafe, after many thousands of lines.

That illustrates the level of understanding of those who argue that curl should be rewritten in rust. Bravo.

[steveklabnik]

Why, specifically?

[acqq]

It's self evident. One personal and very small example (many thousands is obviously less than 10 thousand) doesn't prove anything.

curl has an order of 120000 non-empty lines in c and h files.

The person who made statement of "many thousand" probably made toy applications, compared to something like curl.

More realistic would be to compare something that should have been made in curl: like the "ring" library you mentioned as a Rust answer to "openssl." Which is neither an openssl nor written in Rust.

It is far from providing the functionality of openssl. The author states that:

"ring exposes a Rust API and is written in a hybrid of Rust, C, and assembly language."

"ring is focused on general-purpose cryptography. WebPKI X.509 certificate validation is done in the webpki project, which is built on top of ring. Also, multiple groups are working on implementations of cryptographic protocols like TLS, SSH, and DNSSEC on top of ring."

Actually a wrapper, and actually much smaller than openssl.

Second, "ring" has just some 9000 lines in rs files but 13000 lines of C (as it is defined by the author to be a wrapper, not surprising). So it is still just a wrapper around the C, still orders of magnitude smaller than curl, and still unable to be actually written in Rust.

On that level, I guess the Rust people would call curl the Rust application as soon as a few .rs files would appear in curl code base. Or not.

It's simply has no sense, supporting Rust with misinformation instead of being honest about its current limitations regarding what's actually implemented and what actually doesn't use "unsafe."

It's obvious that those who suggest to others to rewrite the big projects in Rust should first show that they managed to make some much smaller safe code. Especially that they managed to make some meaningful full-Rust and safe libraries.

[steveklabnik]

See, this is much better of a comment than the above, which was mostly an insult, with no substance.

(I don't agree with much of this but that's not the point; my point was trying to raise the level of discourse around here.)