Hacker News new | ask | show | jobs
by luchs 3403 days ago
Something this API gets right is having a unified interface for both IPv4 and IPv6. With the sockets API, you have to decide for one of them. Changing isn't easy as the constants and structures are all named differently.

While it's possible to use IPv6 sockets for IPv4 connections, this doesn't cover all use-cases. For example, you can't do IPv4 broadcast with an IPv6 socket. Additionally, as most examples are written for the classic IPv4 API, that's what everyone uses per default. Later on, when people complain about missing IPv6 support, they are turned down because it's a ton of work to change.
2 comments
dfox 3403 days ago
For majority of applications, supporting IPv6 boils down to using getaddrinfo()/getnameinfo() instead of gethostbyname()/gethostbyaddr(), which results in code that supports both IPv4 and IPv6 and is simpler than the IPv4-only original.
link
rwmj 3402 days ago
The catch is you also have to be prepared to listen on multiple sockets. There are lots of servers out there which get this wrong, including qemu-nbd.

For reference here's my implementation of this which (I think) gets it right: https://github.com/libguestfs/nbdkit/blob/master/src/sockets...

link
mvkg 3402 days ago
A good dual-stack implementation should also support Happy Eyeballs which does not result in simpler code.
link
slrz 3403 days ago
Changing isn't easy as the constants and structures are all named differently.

They are all named "struct sockaddr_storage", no?

You won't even need that most of the time, as it's all dealt with by getaddrinfo and friends (see sibling comment). Sure, there are some cases where it breaks down but the vast majority of software doesn't go there.

link