Aleksa Sarai

Okay, I believe [this documentation example](https://docs.rs/gpgme/0.7.0/gpgme/context/struct.Context.html#method.encrypt) should be updated since that's what I based my code off of. Would you like a PR to update that?

I have considered adding code to `umoci` to do some file locking on `index.json` to try to fix this problem, so if we go in that direction maybe we should...

After looking at this quite a bit, I've become convinced it's actually a `casync` bug -- they aren't generating `mtree`-compatible output. I've opened systemd/casync#167 with proposals for fixing it and...

Yes, sorry. I forgot that `Extra` was implemented (though I'll probably revamp it after #48, which IMO has a nicer interface for differences between manifests).

It's possible that my changes to `dhCreator` haven't handled something properly -- though I doubt it because `/set type=file` still appears to work in the `tar` case -- not to...

I did think about adding this initially, but the main thing that gave me pause was the following: If we assume the rootfs is untrusted (which is why you'd use...

> I'm basically interested in capability-esque use cases where the dirfd is trusted but the user of it isn't. Yeah, this was an idea that has been brought up a...

I hadn't seen it, I'll take a look. It might let me get rid of some of the more awful parts of the syscall wrapping code I have.

From a quick look, I'm not entirely convinced switching makes a lot of sense. My main concern is that there are certain "bad signs" with what flags get passed by...

> All good comments, though I had more hoped the response would be "Let's patch/fix the openat crate". Don't get me wrong, I would be happy to work on patching...