You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I foresee an issue here: rand_core has optional dependencies on getrandomandstd, and now needs to depend on getrandom/stdonly if both are enabled. I'm not sure how to do that. See here where the same issue came up recently.
This is really a Cargo issue IMO, but something we may have to find a workaround to.
The text was updated successfully, but these errors were encountered:
Is this an issue if the rand_core's std feature unconditionally enables getrandom? That seems to be what it does currently.
Also, #106 doesn't actually change much regarding this. std was already optional for many targets on getrandom (like Windows) before that change. The PR just makes the feature act consistently.
No, the problem is that rand_core uses getrandom's std feature for converting getrandom::Error into Box<dyn std::error::Error>. We want to enable std feature for getrandom only if both getrandom and std features are enabled for rand_core, but unfortunately AFAIK it's impossible to define such relation right now, since std = ["getrandom/std"] will enable getrandom optional dependency. So we either can't enable std feature for getrandom or we always depend on getrandom with an enabled std feature.
But actually it does not matter for rand_core v0.5, since it always enablesgetrandom with an enabled std feature.
#106 makes
std
optional.To quote myself (to create an open issue):
This is really a Cargo issue IMO, but something we may have to find a workaround to.
The text was updated successfully, but these errors were encountered: