Re: [boost] [outcome] expected vs. result

Subject: Re: [boost] [outcome] expected vs. result
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2017-05-24 14:44:53

> I am trying to summarize differences between `expected<T, error_code>` and
> `result<T>`. So far, I have identified the following:

Sorry I didn't answer this earlier. I've been trying to reply to the
short emails and bug reports early and deal with the longer replies on a
first received first replied basis. I missed this one.

> 1. API/ABI stability guarantees.

Yes.

> 2. `result` has empty state - `expected` does not.

*Formal* empty state. Yes.

> 3. `result` uses error_code_extended for more error details in the
> ring-buffer.

result and outcome hard code E to error_code_extended. expected defaults
E to error_code. You can, of course, manually supply error_code_extended
as E to expected.

> 4. No ordering (operator<) for `result`

No, nothing gets ordering operators. It is unsafe to combine ordering
operators with implicit conversion according to your own blog!

> Is the above list complete?

No, result has a wide contract on all its member functions such that you
can never achieve undefined behaviour. That's those woolly "default
actions" previously discussed.

expected implements operator*, operator->, .error() etc as reinterpret_cast.

Niall

-- 
ned Productions Limited Consulting
http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/