Josh Stone

> Is there any particular reason why this is private? It's a conservative API choice, not having to worry about how that function looks. Right now it takes `T: AsRef`...

It's going to get _really_ hairy if we try to duplicate cargo's logic for finding all applicable `config` files, as well as parsing the flags from `[build]`, `[target.$triple]`, and even...

Using `cargo rustc` is a clever idea, and it might work to point it at `-p autocfg` to avoid creating a dummy crate. However, from a build script this will...

See also https://github.com/rust-lang/cargo/issues/7501

It can never be foolproof - e.g. `SIGKILL` doesn't give a process any chance to react. But it sure would be nice to handle catchable signals like `SIGINT` (crtl-c), `SIGHUP`,...

Interesting, I didn't know about `chan-signal`. But I agree the application should take control of signals, not the library.

@wfraser nice tip! From libfuse/NEWS, that option was added in 2.9.0, almost 4 years ago. AFAICS this requires no kernel support -- the spawned `fusermount` just daemonizes and waits for...

It's because `thread_scoped` is unsafe, which is because a leaked guard may lead to invalid memory accesses. Same reason that API was deprecated from the standard library. I'm not sure...

In this case, a leak is probably OK, but it deserves careful analysis. The general problem in the standard library involved something like passing a local reference into a scoped...

I was wrong, leaking the guard in this case could be bad too. We're sending the caller's `Filesystem` to a new thread, but since this is generic we don't know...