[muzani]

I think it's fine to make users wait, or even give them a poor experience. I usually go for things were users have already hacked together a solution; the alternative is that they keep hacking even if they'd rather just outsource the work to someone else. If you're working in a field where plenty of competitors have IPv6 support, then copying isn't really your main problem.

You could simply release it first, then add support later when there are a lot of complaints on it. My preferred approach is to avoid paid marketing until there are no more complaints, but launching should happen early to collect complaints.