I have this all the time. Any new rust project and you have to wade through a bunch of once-great crates.
But that's because rust is new. The initial surge over produced solutions to, say binary serialization, and under produced, say, good geodesy libraries. And many many were abandoned. Go to any of the "are we X yet" sites and you'll see many crates that are clearly not finished or advancing which were recently considered SoA.
I've complained about that in the 3D graphics area. But that's niche. Recently I had to write some webcrap, a web server responder that runs under Apache mod_fcgid. I expected that area to Just Work. Works fine in Go, out of the box. There were about a half-dozen abandoned crates which sort of implemented that kind of server. I had to write one from scratch.[1]
The Rust compiler is great, but the crates... There are now enough of them to need curation. Some need to be moved to some obsolete status. Some need to make it to 1.0. Crates.io is starting to clog up with junk.
[1] https://github.com/John-Nagle/maptools/blob/main/server/src/...
The problem is that sometimes library may need to pin a dependency version. Or a dependency was released with a newer major version update and do not back-port security fixes to older versions.
So one cannot just use an old library. Its dependency list must carefully considered.
Now this problem exists with any package management system. But in Rust it is more visible as the language still evolves quickly with non-trivial new features released often.
Then the library authors may want to use newer language features on their API. Then they simply bump the library mayor version and maintain only that. So an old dependencies will not get updates.
We on the Cargo team have been working to educate people on the problems with pinning in Cargo.toml instead of relying on Cargo.lock
> Then the library authors may want to use newer language features on their API. Then they simply bump the library mayor version and maintain only that. So an old dependencies will not get updates.
Thankfully, the ecosystem has mostly settled on build requirements not being subject to SemVer and bump Rust versions in compatible releases. There are a few hold outs.
I think this is a peculiarity of Clojure. Clojure is optimized for simplicity and as the saying goes: "It seems that perfection is attained not when there is nothing more to add, but when there is nothing more to remove."
I don't use Clojure these days. Maybe I should revisit the language. It has some really nice ideas.