r? @huonw

(rust_highfive has picked a reviewer for you, use r? to override)

I think adding PoisonError::new is probably fine, but can this avoid exposing map_lock_result for now? That should be implement-able with new exposed, right?

cc @aturon, an interesting conventions dilemma of whether we should provide constructors for all errors or largely just some method to reconstruct "wrapper errors" after decomposing them.

can this avoid exposing map_lock_result for now? That should be implement-able with new exposed, right?

Yeah, I could copy it out, but I don't want to ;)

As an API we'll have to support to the end of time the threshold should be fairly high for inclusion. A bare map_lock_result doesn't really fit into any conventions that we have so far, so it should likely be left out until a later date if it is necessary to re-include.

As an API we'll have to support to the end of time

It's marked as #[unstable]. I thought we could add things like this as part of the "stability not stagnation".

It's marked as #[unstable]. I thought we could add things like this as part of the "stability not stagnation".

Although we have a distinction of #[unstable] APIs in the standard library it does not mean that it will become a "dumping ground" for everything submitted. The lifetime of a new API is:

1. An RFC is created for a new API and is discussed.
2. The RFC is accepted.
3. The RFC is implemented with #[unstable] APIs
4. Some time later, the APIs are removed or promoted to #[stable]

All currently #[unstable] APIs are either planned for removal or planned to be revisited at some point in the future to stabilize.

Along these lines, I do not think that this map_lock_result is an API that we would want to stabilize, so it should not be included at this time. The PoisonError::new API is a fine candidate for stabilization, however, as it follows a well accepted pattern of constructors and is a very minor (and probably expected) addition to the API.

r? @alexcrichton

(FWIW, I agree with @alexcrichton here.)

That sounds good, I'll update this PR soon.

Updated to keep map_lock_result private.

@bors: r+ 7324c2c

⌛ Testing commit 7324c2c with merge d4f9ec5...

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-win-32-nopt-t, auto-win-32-opt, auto-win-64-nopt-t, auto-win-64-opt