BUG: restore and restoreReplace not using correct native type for the TTL parameter

Expected behavior

The restore* group of functions use int as the TTL parameter, this should actually be long because the redis restore command sets the TTL as milliseconds NOT seconds.

As per the redis docs: https://redis.io/commands/restore
"If ttl is 0 the key is created without any expire, otherwise the specified expire time (in milliseconds) is set".

Actual behavior

Integer overflow when passing a large enough TTL as milliseconds

Steps to reproduce:

String key = "testkey";
int ttl = (int)TimeUnit.DAYS.toSeconds(30); // 30 days as seconds
jedis.set(key, "test-value");
jedis.expire(key, ttl);

// dump record and TTL as milliseconds
byte[] dump = jedis.dump(key);
long pttl = jedis.pttl(key);
jedis.restoreReplace(key, Math.toIntExact(pttl), dump); // results in: ArithmeticException: integer overflow
//jedis.restoreReplace(key, (int)pttl, dump); // unsafe cast results in: JedisDataException: ERR Invalid TTL value, must be >= 0

Redis / Jedis Configuration

Jedis version:

latest

Redis version:

6.0.9

Java version:

openjdk version "11.0.9.1"

[sazzad16]

@NeutrinoAPI Thank you for reporting it.

[walles]

Duration would probably be a better choice than long for durations.