So basically what it comes down to is the argument for building it in Rust is there are some excited people.
And is funny you talk about community, since as the guys on slashdot and others pointed out, the community of people who can use and work in C effectively is orders of magnitudes more than Rust. Have you any idea how many Linux Kernel developers there are alone?
> the community of people who can use and work in C effectively is orders of magnitudes more than Rust
The key word here is "effectively." For the purposes at hand, effectiveness includes memory-safety. Do you know of a single project that implements cmsg(3) in C or C++ in a memory-safe, well-typed, cross-platform way? Or a project where I can submit a pull request and expect it to be reviewed, tested across platforms, and fixed?
I do genuinely believe that the community of people who can use and work in C effectively, in the sense of effectiveness that I and the Tor Project are interested in, is orders of magnitudes smaller than Rust.
I have a very good idea of how many Linux kernel developers there are - and also how many high-severity security bugs there are. I'm a coauthor of a research paper where we wanted to talk about exploitable security bugs in Linux, so we sat down and found a local privilege escalation in hours (CVE-2009-0024).
And is funny you talk about community, since as the guys on slashdot and others pointed out, the community of people who can use and work in C effectively is orders of magnitudes more than Rust. Have you any idea how many Linux Kernel developers there are alone?
http://m.slashdot.org/story/324469
Building API's specific to the use case is part of our jobs as professional developers.
Abstracting a few things things isn't "rewriting libc", that is just general practice for most decent size projects.
Anyway, the decision is made, so the whole thing is moot at this point.