Document the default for `ExitStatus`

[jyn514]

@scottmcm: I think it should at least say somewhere what that default is. Having it come from a derive(Default) on a c_int feels indirect to me as well, so perhaps it would make sense to define it manually in terms of ExitCode::SUCCESS instead, say -- that would also allow putting a doc-comment on the impl.

We should document that Default for ExitStatus is "process reported successful termination", ie the result of ExitCode::success().into(), and, implement it that way.

Originally posted by @ijackson in #106425 (comment)

[john-h-k]

Is this possible to currently do? Having checked out the linked issue it looks like it depends on impl Default for ExitStatus existing

[jyn514]

Right, it needs #106425 to merge first.

[john-h-k]

Alright 👍 . How does documenting work for a derived trait?

[jyn514]

You'd have to remove the derive and use manual impl Default blocks.

[ijackson]

You'd have to remove the derive and use manual impl Default blocks.

That ends up with the docs for what Default means lower in the rustdoc page, next to the place where the impl and its methods appear. It might be better to just document it in the doc comment for the struct.

[john-h-k]

Also just to check, It's ExitCode::SUCCESS.into() right? success() doesn't seem to be a method and doesn't seem to be being added by that issue (or should I add that method?)

Also, should I open a PR (that is on top of the existing PR to actually add the default), or wait until the other is merged? Done the changes locally already

[ijackson]

Also just to check, It's ExitCode::SUCCESS.into() right? success() doesn't seem to be a method and doesn't seem to be being added by that issue (or should I add that method?)

Sorry, I think that was me writing loosely. Yes, I mean te constant SUCCESS I thijnk. You don't need to add anything. Ian.

[u-na-gi]

@rustbot claim

[ijackson]

Note #114593, which is a bit of a swamp. I think changing the docs can be done (at the cost of adding to the set of things which the docs claim to be true but which are false on those broken platforms) but changing the implementation ("and implement it that way") is blocked.

[john-h-k]

@rustbot claim

I’ve already got a branch with this issue fixed on it - was waiting for the previous PR to merge and it now has. For some reason I forget to claim it though