Rust needs RedHat and the like

For Rust to become famous like Java, Python, etc...there is a need to have some big company like RedHat to officially publish and maintain a set of enterprise grade crates. Crates that companies can trust and be sure there is support if ever needed. For the lanugage itself, there is also a need for a clear vision of future releases in a calendar.

There are way too many dead crates that make Rust look like a non enterprise production language. I was looking at crates for ZeroMQ, and this is more like a cemetery. This is not the first time, this is just an example...

CTO will never adopt Rust if somebody doesn't drive that thing officially and seriously with commercial support, license model, support etc...

I beg to differ. Microsoft is using rust quite intensively, so is Google. Facebook implements Libra in Rust, Cloudfare uses it, and many more.

There is a few things that could help move forward. For example there is no official language reference. The compiler implementation is the reference which isn't great.

Right now rust is a moving target, since it's still so young and it is so actively developped. This problem will probably solve itself over time as it becomes more feature complete.

As for the ecosystem, yes, high quality reference crates are still missing in many domains, but it's a matter of time before these arrive. Do you think the 700k+ packages in npm are no cemetery?

5 Likes

I am a CTO. All our new development is in Rust. Except where we need to use Python.

Mind you, we are a very small outfit, a drop in the ocean.

On the other hand, in a former life working for various companies, including Nokia, none of the projects I worked on since about 1998 have depended on commercial support, licence models etc. We just grab Linux, GCC, whatever opensource software we need and get on with it.

In fact for one major project we switched from a commercial operating system to Linux, mid-development, because the commercial offering was buggy and support was terrible.

Sure there may be a lot of dross in the world of Rust crates. Much like there is in the world of NPM for Javascript. That did not stop the likes of JS or Python becoming "famous".

What would make me feel totally secure in the choice of Rust would be multiple/alternative compiler vendors. A Rust for GCC, and MS Rust and so on.

3 Likes

What you wish takes too much manpower.

Rust does not have a runtime so crates (and compiler) stay with the developer and do not have to be passed onto users.

Rust has cargo audit. It should be a enterprises first point of call to keep software safe.

History shows lots of closed environments do not survive. (think sourceforge vs github.) The unmaintained and incomplete is a small consequence of being open.

The "clear vision" for enterprise is the compiler will be (mostly) backwards compatible with 1.0 crates. The language has Editions to allow new developments to go ahead with otherwise incompatible functionality. There will never be a reliable calendar for future feature releases, and enterprises should not depend on such a thing. e.g. It was hoped async would come with the edition but was a year after before stability.

2 Likes

I work in Red Hat's toolchain team to provide the Rust compiler, with the same support as any other package we provide. I also maintain and contribute to many other crates, but those are not directly provided in Red Hat products. We do have supported products using Rust, like Stratis Storage and CoreOS, and those have published crate components too.

2 Likes