People aren't building castles on foundations of sand. They are building sandcastles, experimenting and building along as the language and ecosystem evolves and matures. Every nightly-only crate should be regarded as highly unstable and experimental. If a lot of people are building sandcastles surely that means that the language wasn't yet ready for that particular castle being made of stone.
And this is a good thing! AFAICT without nightly-only crates being so pervasive due to compiler plugins, Macros 1.1 wouldn't be so close to stabilization. Isn't that exactly what nightly is for? If some nightly feature is used a lot, that means it should promptly be polished and stabilized.
At such an early stage in the language's life the evolution is mutual. The development team responding timely to such issues is just another sign that Rust is a healthy (even if in its infancy) language.
Stable Rust is dead, long live Stable Rust!
> Stable Rust is dead. Nightly Rust is the only Rust.
Many people and orgs are using Rust stable. In production.
Nightly is a playground.
We shouldn't misinform those unfamiliar with the language or give them the opinion that Rust isn't well managed. Doing so can steer new talent and interest away. We are all relying on newcomers for more widespread adoption and faster growth of our community.
The author probably didn't mean any harm, but I feel like the article is disingenuous and should have a retraction before it causes any confusion.
(And for the record, I think Rocket is the perfect sort of project for helping us get experience with using unstable features, and helping us prioritize which should be stabilized first. I don't want to knock it just for using unstable features! People are too eager to forget that, as of two years ago, every Rust project was de facto unstable, and we've been making steady progress toward a majority-stable ecosystem since then, as expected.)