proposal: net: add IsCGNAT to the net.IP type

[thomasmarton]

Proposal Details

The standard library for Go already provides a ton of convenience methods for developers looking to filter out for specific IP addresses. For example IsPrivate checks if the IP belongs to a private network defined in RFC1918 or IsLoopback that checks if the IP address is a loopback address.

I'd like to propose a similar method called IsCGNAT that checks if the IP belongs to the Carrier Grade NAT space defined by RFC 6598.

The method would return a bool just like all the other helper methods and to avoid changing the current API conventions, it would return false for any IPv6 address provided as there is no current RFC for an IPv6 based CGNAT.

[gabyhelp]

Related Issues

• net: add IP.IsPrivate (RFC 1918 & RFC 4193) #29146 (closed)
• net: apply ICANN/IANA-managed semantics to IP.IsGlobalUnicast #11772
• proposal: net/netip: add Unmap() to netip.Is4() #49970 (closed)
• proposal: net/netip: function to split ip range to into cidr networks #53370 (closed)

Related Discussions

• proposal: net/netaddr: add new IP address type, netaddr package (discussion)

_{(Emoji vote if this was helpful or unhelpful; more detailed feedback welcome in this discussion.)}

[seankhliao]

Why do we need another method?
What existing code would have benefited from this?

IsPrivate was added only because it was a series of ranges, we rejected any other special ranges in #29146

[thomasmarton]

First of all my apologies for missing that issue, I tried my best to search around for proposals and also issues that might mention it.

From how I understand #29146, it simply wants to avoid unnecessarily cluttering the struct with similar methods that might not be necessary. I feel like with the widespread adoption of tools such as Tailscale, Netbird and all kinds of other WireGuard mesh network solutions (that almost always use the CGNAT address space for addressing members) the inclusion of this additional method might be useful.

I personally would not want to change IsPrivate to include this range as that method is strictly aligned with RFC 1918 but I think this method would be worthy of consideration given how some private networks now use this IP range rather than the conventional private ranges defined in RFC 1918.

[seankhliao]

based on some loose searches

https://github.com/search?q=language%3AGo+%2Fiscgn%3Fa%3Ft%3F.*net.ip%2F&type=code

https://github.com/search?q=language%3AGo%20%2F100.64.0.0%2F&type=code

https://github.com/search?q=language%3AGo+%2F100%2C+64%2C+0%2C+0%2F&type=code

https://github.com/search?q=language%3AGo+%2F%28ip%7Caddr%29.*%5C.IsPrivate%5C%28%5C%29%2F&type=code&p=1

it doesn't appear to be used or defined that often, and more importantly, it doesn't seem as if having a method would have been useful to most of the existing code that has needed to write out the range

[neild]

From what I've seen, IsPrivate was a mistake. Just about every use of it that I've seen seems to misuse it. For example, declaring that "if net.IP.IsPrivate, then the host at address is trusted", which is generally not a safe assumption.

net/IP.IsCGNAT can report whether an address is in the RFC 6598 range, but it can't report whether an address is being used for CGNAT, since some carriers use space outside this range for NAT. So the question I have is whether the small convenience of a method outweighs the confusion caused to users.