arma said I might have an opinion, and he was right:
I think the Tor Browser default means that tor has more users with these options on than off.
If it works for Tor Browser, I can't see the harm in turning it on.
But I'd like to know what consensus weight of the total exit consensus weight supports IPv6.
(My stem isn't quite up to the task.)
(One way of increasing this proportion is to have exits automatically use IPv6 when it's available. I think there's a separate ticket for that.)
the default-setting code for these options is duplicated in at least 3 places. There's a ticket to fix that, too.
I can only open https://ipv6.google.com/ in Tor Browser 1/10 times when using "New Circuit For This Site", so maybe we don't have enough IPv6 exits for PreferIPv6 yet. That said, if IPv6Traffic is off, that would be 0/10 times.
I'm using the attached patch, i think it passes the tests, and i can confirm that situations that were broken before for me (connecting to v6 hosts through a stock tor installation) are now working with this patch applied.
But I'd like to know what consensus weight of the total exit consensus weight supports IPv6.
As of onionoo data from 2017-01-23 20:00,
12.57% of exit_probability supports exiting to IPv6 according to their exit policy. I could add it to ornetstats if regularly updated data is needed.
(One way of increasing this proportion is to have exits automatically use IPv6 when it's available. I think there's a separate ticket for that.)
I second teor's here opinion on PreferIPv6, I fear we don't have enough Exit on IPv6 so if we get this patch in 030, let's advertise this a lot that we need Exit to start enabling IPv6 more and more! :)
I've taken dkg's patch, added a changes file and put it in branch ticket21269_030_01 (authorship has been kept).
lgtm and I confirm it's working as well! Putting that one in merge_ready so if we think it's a good idea, I think it's ready for nickm's eyes.