From: roundtable-bounces@muug.mb.ca [mailto:roundtable- bounces@muug.mb.ca] On Behalf Of Mark Jenkins
I've been playing with IPv6 for a few years now, recently participated as both an end user and service provider on IPv6 day (June 8). Have recently been thinking about putting myself on an IPv6 only diet at both work and home to try and break my addiction to news and articles on the web so I can get some real work done. (most of my favourite time killer web sites don't have ipv6 enabled; the AAAA records are not even available in the name server my tunnel broker uses which does successfully query records unavailable to most like google.ca AAAA)
FYI, one of the FreeBSD developers attempted to go IPv4-less, and did a presentation about it at BSDCan this year.
Short version: the stack is ready, the apps REALLY, BADLY aren't ready.
"Why do apps matter?", you wonder... any code that uses struct inetaddr is automatically IPv4-only. Any code that calls gethostbyname() et al. is IPv4-only (because that uses struct inetaddr). Then there's the truly weird bugs that crop up in otherwise-IPv6-enabled apps when they don't have a valid PF_INET to fall back upon.
There's a link to the audio recording of Bjorn's talk at http://www.bsdcan.org/2011/schedule/events/222.en.html.
Given that most of his problems were app-related, I can't see Linux being any better. Nor, for that matter, Windows or Mac OS X - at least at this point in time.
(FWIW, he wasn't any easier to understand in person...)
-Adam Thompson athompso@athompso.net