Can't compile 0.7.0 in no_std environment

[russellmcc]

Starting from cargo new rand_test, I added the following line to Cargo.toml: rand = { version = "0.7.0", default-features = false }. Then, replaced main.rs with the following file:

#![no_std]

fn main() {}

Then, I tried to compile it for arm with cargo build --target thumbv7em-none-eabihf.

I expected errors about not including exception handlers/runtime, but instead, I got the following error:

error[E0463]: can't find crate for `std`
 --> /Users/russruss/.cargo/registry/src/git.luolix.top-1ecc6299db9ec823/lazy_static-1.3.0/src/inline_lazy.rs:9:1
  |
9 | extern crate std;
  | ^^^^^^^^^^^^^^^^^ can't find crate
  |
  = note: the `thumbv7em-none-eabihf` target may not be installed

Is there a different set of feature-flags I need to disable std support in rand?

[russellmcc]

The same test with 0.6.0 seems to work great.

[KodrAus]

It looks like somewhere in our dependency graph is a version of lazy-static that isn't disabling its std support.

If you look in your Cargo.lock are you able to tell which dependencies are pulling in lazy-static?

[russellmcc]

As I mentioned, only rand was directly added to my dependencies. Here's the output of cargo-tree:

└── rand v0.7.0
    ├── libc v0.2.62
    ├── rand_chacha v0.2.1
    │   ├── c2-chacha v0.2.2
    │   │   ├── lazy_static v1.3.0
    │   │   └── ppv-lite86 v0.2.5
    │   └── rand_core v0.5.0
    └── rand_core v0.5.0 (*)
    [dev-dependencies]
    └── rand_hc v0.2.0
        └── rand_core v0.5.0 (*)

So, it looks like c2-chacha might have something to do with it - I'm not sure I know a tool that can dump out which features are enabled in the dependency graph.

[vks]

If you build rand without the std feature, this should propagate to c2-chacha, which should build without lazy-static at all. I'm puzzled why this is not the case.

[vks]

If you run cargo tree --no-default-features on the Rand root, lazy-static does not appear in the dependency chain. However, for a project depending on rand without default features you get the output above. Is this a Cargo bug?

[burdges]

Isn't that the cargo bug where crates wind up with their test features?

[vks]

@burdges You are right, it seems that the dev-dependency rand_hc pulls in std somehow.

[russellmcc]

Aha; it looks like this is fixed on the master branch. Sorry for the noise! Eagerly awaiting a new release!

[dhardy]

Sorry, I have limited access at the moment, but if @vks opens a PR preparing a new release I can do the rest. However maybe we want to deal with some of the other PRs first?