Address failing tests

[gillchristian]

TTL race condition?

Setting expiration on a key and checking ttl on it right away produces different results.

The following commands:

SET key 1 EX 1
GETEX key EX 10
TTL key

Return a ttl of 10 on Redis and 9 on our server.

It isn't clear if this is a race condition on the tests (I doubt it) or if our TTL implementation works differently.

Keys sorting

Introduced sorting by insertion time to our keys command implementation to make it deterministic.

Redis' seems to be deterministic (always the same for a given set of keys) but not guaranteed (changes across different runs and instances, eg. local vs. CI tests).

I changed the tests to only return one key at a time.

[ndelvalle]

I wonder if we should not update the implementation but rather the tests, doing a comparison that does not care about the orden.

[ndelvalle]

created_at? I've always seen/used created_at and updated_at on db records.

[gillchristian]

Yeah, that's better

[gillchristian]

I wonder if we should not update the implementation but rather the tests, doing a comparison that does not care about the orden.

Even if we update the tests, I'd still keep the deterministic order on the implementation.