getrandom: rand_core and error types

If we are to make this crate independent of rand_core, then we need to include an error type here.

We could simply copy the one rand_core uses; I think something simpler and equivalent on no_std may be preferable however. A quick look at the code shows that where we do include a cause, we are mostly just using an integer error code. Whether it is even worth forwarding the cause is another question.

About this issue

Original URL
State: closed
Created 5 years ago
Comments: 15 (15 by maintainers)

Most upvoted comments

I don’t think that such extra-complexity is warranted (plus it will be even less idiomatic that NonZeroU32). On panics users by default already will get an error message specifying that it originated in getrandom. After that they’ll have to look into platform specific details either way.

As was written in the rust-random/rand#715 I think in addition to msg we could add is_retryable method and RETRYABLE error constant. Also I am not sure if automatically retrying on Interrupted error is a correct behavior. For example on Linux interrupt handler could use SA_RESTART flag to continue interruptible operations without EINTR, so hard-coding retry loop may be a wrong approach.

newpavlov on Mar 11, 2019

There is an option we could use here if we wanted: something akin to the UNIX errno, but using an AtomicPtr to store a &'static str, behind an API like

fn set_err(msg: &'static str);
fn get_last_err() -> &str;
// or, for cfg(std):
fn set_err<T: ToString>(msg: T);

This would be fully no_std compatible with minimal run-time overhead (on successful usage). There are two drawbacks:

some extra code size / memory usage
a more complex API, which may be a problem for inclusion in std; also we would essentially be piggy-backing this crate to define an Error type possibly used elsewhere in Rand

dhardy on Mar 7, 2019