> The whole rust nightly/beta/stable hasnt worked well, because (afaik) the beta usage is tiny and people tend to jump on nightly or stable. It’s only very recently any effort has been made to change this for the new edition.
My impression was that most people feel the Beta toolchain is doing its job just fine, because most projects that use CI include a Beta test run. I don't think the plan was ever to have people using Beta on the command line as their daily driver.
Can you give an example of soundness regressions in the previous release? Neither the 1.29.1 release nor 1.29.2 are indicative of people not using the beta; the former concerned a potential security vulnerability in a stdlib API that had existed for several releases, and the latter concerned a runtime heisenbug that could neither be found by merely compiling with beta nor even reliably by running the code.
> Notice specially the last month or so of actively asking people to try the beta.
This is because for the last month the beta has represented the first release candidate for Rust 2018, which is more important than normal releases in all three of implementation scope, backwards compatibility concerns, and marketing potential.
My impression was that most people feel the Beta toolchain is doing its job just fine, because most projects that use CI include a Beta test run. I don't think the plan was ever to have people using Beta on the command line as their daily driver.