Julien Cretin

If permissions to persistent storage are defined in the application header, then application IDs are not required, so we can resume the discussion here. Let me recap the current state...

Note that there's the same issue with `cargo.noDefaultFeatures` which doesn't work in a cargo workspace.

Isn't the problem that `SecretBox` should not be implemented as `Secret` but as a custom type instead? The reason is that we want `T: Zeroize` not `Box: Zeroize`. ```rust pub...

Cool! Thanks for the feedback and no worries. Nothing urgent, just wanted to check my understanding.

> I don't think this use of drop is idiomatic That's a fair point. See below for a solution if we go in that direction. > do you have any...

Sounds good to me. I also think that silencing the lint at top-level of match branches (i.e. ` => drop(),` pattern) should be enough indeed.

Regarding storage, the proposed implementation is a valid idea (adding a file-based `Storage` implementation). I'm not sure why the `BufferStorage` is used though, since the file is probably going to...

I see. That's actually a good reason to change the API then. I was considering it for other reasons (in particular make it compatible with `embedded-storage`), but this one adds...

This issue is not about a problem in the continuous integration setup nor in the test itself. Both are working fine, although they could be improved as you describe. This...

> So I think pinning a Cargo.lock and a compiler version Yes, we should do that. > an extra workflow running as a cron job to notify when dependencies are...