cannot tell if meaningful much less true,
my brain isn't very computersingitizededu
a word I made up
just to set a rhyme at you
*clucks like a chicken three times*
@lanodan The point is not to "not use C" the point is to reduce the amount of unsafe code that we depend upon to be safe. The input to LLVM is generally considered trusted, the input to a rust web service isn't.
It's already been shown that even well established code still has safety issues. (sudo, for one).
Yes, *all* changes to a codebase have the chance of introducing bugs. You can incrementally introduce rust into a project to minimise issues, like what curl is doing.
Rustls (and its dependencies) have been audited: https://jbp.io/2020/06/14/rustls-audit.html
that's actually what's going to replace the TLS in Apache. mod_ssl is C (presumably), but mod_tls uses rustls instead of openssl.
Yes, ring does have C code in it (looks like it's from the BoringSSL project). Unfortunate, but pure rust crypto does exist. The point is to keep unsafe to a minimum, so instead of needing to audit 100% of a codebase for memory safety bugs, you need to audit 5%.
@132ikl I find this hilarious! 😂
I mean I think it's pretty obvious it's meant to be in jest, but it's funny because it's true.
@132ikl I don't disagree with the sentiments, but I'm not sure you can extrapolate what we know today to what was known about in the early 1970s when C came about - it feels a bit like hindsight.
I mean, it was meant as an improvement to BCPL (or, a variant of it) - and to be fair, it was an improvement on BCPL and its variant, it's just that we know a lot more nowadays.
@132ikl "Wanted to access memory directly anyway? We had a tool for that: It was called "ASSEMBLY"" is literally us with libseven
A Mastodon instance for programming language theorists and mathematicians. Or just anyone who wants to hang out.