Replaced the accessors that return optional string_view objects with two separate functions.

[glynos]

This is what we talked about earlier (or last night, if you are reading about this when you wake up).

I removed the use of optional, and updated all the tests. For me, all tests are green. So the following is up for discussion:

1. Do we need "bool has_x() const" if string_view::data() can be tested against nullptr;
2. What do we do with "template intT port() const"?

For 1, I think (weakly) that the answer is "yes"
For 2, I think throw an exception on error

I'm already comfortable with the idea of no longer using optional in this context, my experience already is that is much easier for the user to use and understand.

[glynos]

P.S. If we agree, I want to start integrating this (as a submodule) into the cpp-netlib

[deanberris]

Do we need "bool has_x() const" if string_view::data() can be tested against nullptr;

I think so. It's useful to have these specific checks.

What do we do with "template intT port() const"?

I think we can turn this into a string_view too. If users want to turn this into a number, they can choose the many ways to do it.

I'm already comfortable with the idea of no longer using optional in this context, my experience already is that is much easier for the user to use and understand.

SGTM

[deanberris]

These probably should be noexcept alongside const.

[deanberris]

LGTM