Presumably they'll have to sort out what kind of performance hit it has and how to address that longterm, but I think you're correct.

It's probably more complicated than that given that the Redis security page begins with:

> Redis is designed to be accessed by trusted clients inside trusted environments.

https://redis.io/topics/security

I think the current recommended approach for this use case is to wrap it with stunnel.

Please re-read. It's spiped. Fixed 1k blocks, but otherwise reasonable.

But #4855 is going to be merged, and my tests show that #4855 is actually faster than my changes, so ... :)

I like not having to be responsible for writing SSL/TLS code, even if I just spent 9 weeks doing it. :P

It looks like Amazon's patches are already on their way in: https://github.com/antirez/redis/pull/4855

I'm curious too. Why fork for this one thing? There must be a good reason.

Not just one thing, multiple things. Transactions, performance improvements, and long-term, more, that rely on Transactions and other parts that Salvatore has expressly stated that he wasn't interested in.