IPv6 host normalization.

[ioquatix]

I reviewed the PR #1213 and it looks like we recently merged #1538 which prefered the non-square-brackets representation. However, this has not been released yet. So we could adjust it.

Based on the defintion of URI#host, I think we should probably prefer the square brackets representation... but I'm not sure what is canonical. I need to do some research.

Originally posted by @ioquatix in #1213 (comment)

[ioquatix]

@jeremyevans said:

RFC 7239 Section 7.4:

   Note that IPv6 addresses may not be quoted in
   X-Forwarded-For and may not be enclosed by square brackets, but they
   are quoted and enclosed in square brackets in "Forwarded".

       X-Forwarded-For: 192.0.2.43, 2001:db8:cafe::17

   becomes:

       Forwarded: for=192.0.2.43, for="[2001:db8:cafe::17]"

So I think we should handle this for X-Forwarded-*. but not Forwarded (when support is added
for Forwarded #1423).

To me, Forwarded header is more recent, and it prefers quotes.

Additionally, a URI#host field is quoted:

URI.parse("http://[2001:db8:cafe::17]:47011/").host
=> "[2001:db8:cafe::17]"

That being said, you can't feed such a value to Addrinfo.tcp. Nor do command line tools prefer it:

koyoko% ping "[2406:e003:1d0f:fa00:ec4:7aff:feb5:9500]"
ping: [2406:e003:1d0f:fa00:ec4:7aff:feb5:9500]: Name or service not known
koyoko% ping "2406:e003:1d0f:fa00:ec4:7aff:feb5:9500" 
PING 2406:e003:1d0f:fa00:ec4:7aff:feb5:9500(2406:e003:1d0f:fa00:ec4:7aff:feb5:9500) 56 data bytes
64 bytes from 2406:e003:1d0f:fa00:ec4:7aff:feb5:9500: icmp_seq=1 ttl=61 time=4.45 ms

On https://bugs.ruby-lang.org/issues/14016 it states:

Use URI::HTTP.hostname which unwraps the bracket.

u = URI("http://[::1]/bar")
p u.hostname      #=> "::1"
p u.host          #=> "[::1]"

So, maybe we should adopt this model.

[ioquatix]

@AlexWayfer do you have any opinion on this?

[AlexWayfer]

@AlexWayfer do you have any opinion on this?

To be honest, no, because as I wrote in #1538:

I'm not sure about IPv6 format, you can correct tests and I'll adopt the code.

I'm not familiar with IPv6 standards and formats.

[ioquatix]

I want to follow the same naming conventions as URI#hostname and URI#host, along with the canonical definitions of the authority header in HTTP/2. I am working on a PR but there are a couple of tricky edge cases!

[AlexWayfer]

I want to follow the same naming conventions as URI#hostname and URI#host, along with the canonical definitions of the authority header in HTTP/2. I am working on a PR but there are a couple of tricky edge cases!

You always can publish draft PR and we can try to help you with them.

[ioquatix]

My goal is to use the following terminology:

• Authority (as per HTTP/2 spec): provided by host header, should be formatted as address:port, ipv4address:port or [ipv6adderss]:port

• Host (as per URI spec): the address part including delimiters, without port.

• Hostname (as per URI spec): the address part excluding delimiters, without port.

[AlexWayfer]

Hostname (as per URI spec): the address part excluding delimiters, without port.

Does "delimeters" mean square parentheses?

I wanted a helper method for host without port or with it (as it in headers) in #1435, and we've chose hostname for this.

[ioquatix]

@AlexWayfer Yeah delimiters means square parentheses.

Yeah, agreed we need some better accessors.

Request#hostname should be the thing you can pass to Addrinfo.tcp.

[ioquatix]

I've reworked the PR, so that the first commit is the failing specs.

[AlexWayfer]

I've reworked the PR, so that the first commit is the failing specs.

I see now specs are passing. 🎉