4.0.0b1 breaks script_flush on pre-6.2 server

[bmerry]

Version: redis-py 4.0.0b1, redis 5.0.7

Platform: Python 3.8.10 on Ubunutu 20.04

Description: In 4.0.0b1, the implementation of script_flush always calls either SCRIPT FLUSH SYNC or SCRIPT FLUSH ASYNC on the server. However, the SYNC/ASYNC modifier was only added in Redis 6.2, making it fail on older versions.

To reproduce (assuming a Redis server on localhost):

import redis
r = redis.Redis('localhost')
r.script_flush()

Result: ResponseError: Unknown subcommand or wrong number of arguments for 'FLUSH'. Try SCRIPT HELP.

Apart from wanting to support older Redis versions, the documentation suggests that providing neither modifier is not simply equivalent to providing SYNC. Rather, it leaves it up to a configuration setting on the server, and I feel like calling r.script_flush() should respect that server-side default.

[chayim]

Thank you for reporting @bmerry. The PR you see referenced above looks to fix this. We now default to None, which solves both behaviours. Of note, the server can have a different default set - meaning this is a better implementation. Thank you for catching this! It'll be in 4.0.0b2!