Fix a number of server compatibility issues

[jgosmann]

• This allows IP addresses like 127.0.0.1 as websocket origin for servers listening on localhost. Actually the whole 127.0.0.0/8 is allowed because all of these addresses are reserved for loopback. This is meant to allow direct use of the IP if localhost does not resolve correctly.
• Adds IPv6 support and allows [::1] (and equivalent addresses) accordingly as websocket origin. This is meant to allow to address the server with localhost on computers where this resolves to the IPv6 address (though with a sane configuration the IPv4 address should be tried when the IPv6 address fails).
• Safari complains when you send LF instead of CRLF, so we're sending CRLF now.

This is all meant to address issues reported in #807. However, one issue remains: Firefox does not support using IPv4 addresses instead of localhost because it complains that accessing localStorage is not a secure operation. I won't fix this for now because it is not related to the server refactoring and I don't understand why exactly Firefox complains (Chrome and Safari are fine with it.)

[mention-bot]

@jgosmann, thanks for your PR! By analyzing the annotation information on this pull request, we identified @tcstewar, @tbekolay and @pabogdan to be potential reviewers

[Seanny123]

This fixes the problem in Chrome, but not Safari, where everything is only half-loaded:

[jgosmann]

What URL are you using?

On July 14, 2016 8:56:00 PM EDT, Sean Aubin notifications@github.com wrote:

This fixes the problem in Chrome, but not Safari, where everything is
only half-loaded:

---

You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub:
#808 (comment)

Sent from my Android device with K-9 Mail. Please excuse my brevity.

[Seanny123]

The localhost one

[jgosmann]

Safari gives me some CSS errors, but seems to work for me.
Are there any suspicious messages in the developer console?

[Seanny123]

I lost access to the computer that was causing the problem, so I can't
follow up on this. Let's ignore that problem for now.

On Thu, Jul 14, 2016 at 6:10 PM, Jan Gosmann notifications@github.com
wrote:

Safari gives me some CSS errors, but seems to work for me.
Are there any suspicious messages in the developer console?

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#808 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AB1HaR0phZQ2eC4GGvz5WwK1ivFgLd7Nks5qVt4XgaJpZM4JM-rt
.

[Seanny123]

Could you add a quick docstring about what this function is doing? From what I can parse this is used to verify if the expected host was reached? This function only seems to get called in assert statements.

[jgosmann]

It would say something like "Checks whether host refers to the loopback interface." Not sure if that really adds any value because to me it is what the function name already says in more words.

[Seanny123]

You're right. I wrote that comment when I was sleepy. I shouldn't review code when I'm sleepy.

[Seanny123]

LGTM

[jgosmann]

This PR has become obsolete.

• The problem of IP addresses as websocket origin has been solved in Token based authentication #915 as it requires a login for any sort of connection even for localhost. Thus, no special case code for that is needed.
• IPv6 support has been added in a better way in Add dual stack IPv6 support. #864.
• A fix for the Safari problem is also already merged into master and released as 0.3.1.