build(dep): allow rand 0.10 in addition to 0.9

[tisonkun]

cc @shepmaster

[shepmaster]

Pretty sure this won't (shouldn't?) work as rand changed their MSRV.

[tisonkun]

Pretty sure this won't (shouldn't?) work as rand changed their MSRV.

Fair enough. Thanks for triggering the CI workflow.

I'm working on bumping rand to 0.10 in fastrace as well and meet the same issue (fast/fastrace#162).

Will twox-hash upgrade MSRV and edition recently or have a plan to do that?

[shepmaster]

Will twox-hash upgrade MSRV

I don't know. I have not yet decided what this crates relation between MSRV and SemVer will be.

Another annoying thing is that in many cases I can just widen the dependency version (e.g. rand = "~0.9, ~0.10") because the specific code paths that I use don't change from version to version. Unfortunately, it looks like rand did change them this time.

One possible solution, like I do over in SNAFU, is to have a feature flag like rust-1.85. Then I think I could use rand 0.9 in not(rust-1.85) and rand 0.10 otherwise.

Will twox-hash upgrade [..] edition

There's not really a strong reason to do so, so I doubt it. Can you expand on how this crate's edition impacts you?

[shepmaster]

Unfortunately, it looks like rand did change them this time.

Oh, that's only in the testing code? That might be fine then. Let me noodle for a bit.

[shepmaster]

I've force pushed to your branch after rebasing it and widening the rand dependency. Doing so has un-verified your commit (as it should).

Once I've got CI passing, please do whatever additional testing you'd like against this new version of the code and then rebase it yourself and re-verify it.