Capability.
Honestly, I expected more. Also from/in Europe and the US. (Numbers mentioned in the article.)
Yeah, I assumed this to be a situation where 90+% of devices are capable, but people don’t switch over out of laziness. That’s kind of insane to me that it’s only at 50%.
Today, at $DAYJOB, we wanted to assign a temporary IP address to the Ethernet port of a Raspberry Pi, so we could connect an Ethernet cable to it and SSH in. We considered using an IPv6 address for that, because there’s a lower chance of someone already using that IP address.
But as far as we could tell (didn’t investigate terribly long), NetworkManager in Linux will kill a connection, if it does not get an IPv4 address in time, because it assumes there to be a DHCP server behind an Ethernet port. So, we decided to just use an IPv4 address instead.
I’m sure there’s ways to make NetworkManager not do that, but those involve additional steps and complexity. And yeah, it’s just kind of ridiculous to me, that we built these workarounds, like DHCP, to make IPv4 work and then the defaults assume those workarounds to be in place, which means the IPv6 experience is worse than the IPv4 experience with its workarounds…
I kind of hate ipv6 just for all the different abbreviated formats.
They’re a pain in to parse, and frankly that’s a burden to developers.
You’re supposed to use standard library functions for parsing IPv6 addresses, like inet_pton
