[mustache_kimono]

> Entire programming languages like Rust depend on this to install the compiler

And it works! For people who have to stay on top of language changes (not everyone), who are on a wonky platform or that doesn't update quickly enough, this is actually a pretty okay method. And they do also offer alternative methods: https://rust-lang.github.io/rustup/installation/other.html

Here, the "You know what they should be doing..." attitude surely doesn't account for something.

> rustc changes too fast for anything but the most rolling of distros to keep up

This may have once been true, but isn't really anymore. Ubuntu updates all distros with the latest and greatest and does so about every couple of minor version releases (last was 1.59 to 1.61). This is fine for me. My MSRV is now whatever Ubuntu is shipping.

[superkuh]

>For people who have to stay on top of language changes (not everyone)

Or anyone that wants to try out the latest "$something but in rust" and tries to compile it.

>who are on a wonky platform or that doesn't update quickly enough,

ie, the vast majority of people so I question the word "wonky" here.

>This may have once been true, but isn't really anymore.

It's still true in my experience. But then again, Rust does change really fast. Maybe they fixed their entire bleeding edge demographic in the last couple weeks and now people refrain from using $latest features that don't work in rustc from 3 months ago.

[mustache_kimono]

> Or anyone that wants to try out the latest "$something but in rust" and tries to compile it.

This is kinda true. I think a smallish demo re: who usually downloads a compiler, but it happens. I had a very sophisticated user/dev file a bug about how my software wouldn't compile. Turns out re: compiling from source they downloaded an old version of rustc from their distro's repos, instead of following my instructions, and my new sources wouldn't compile.

Now this was user error. But it happens. Just re this compiler error it was a good change and an easy fix. I stick to a MSRV now.

> ie, the vast majority of people so I question the word "wonky" here.

The vast majority of people aren't downloading compilers.

> Maybe they fixed their entire bleeding edge demographic in the last couple weeks and now people refrain from using $latest features that don't work in rustc from 3 months ago.

Price of using a non-dead language I'm afraid? You and I don't have to worry about the current patois of Latin or ancient Greek either? Remember -- the problem is I create something new, which compiles with a newer version of the compiler, you download the sources and compile, and you get an error because you have an older version of the compiler. This happens in every ecosystem, whether its C, Python or Rust.

If you're argument to me/the kids is: It should be old and nothing should ever change, I'm not sure that's a winner.

[superkuh]

> It should be old and nothing should ever change

Bash gets new features, forwards incompatible features, fairly regularly still. It changes almost as much as Rust. But guess what? Bash devs don't expect that it's only other Bash devs using the latest bleeding edge version from last month. They, in general, write code that will work anywhere any time.

Rust code could be written in this way. But it it isn't because of the type of demographic that current writes in Rust. I have confidence this will change over the next decade. But right now, fast forwards incompatible changes combined with fast forwards incompatible devs makes for a very limited lifespan of any rustc. And that leads to the vast majority of the documentation suggesting curl | sh. You don't get that with other languages.

I'm not arguing for no change ever. That's silly. I'm arguing for not writing for the bleeding edge just because you can.