The use of EVAL prevents using node-redlock in dev/test environments about node-redlock HOT 2 CLOSED

Hi @natesilva, this is a good question, and you are right that the lock command doesn't really need to be an EVAL. However, the unlock and extend commands really do need to, and given this, I made the lock script an eval for the sake of symmetry and maximum extensibility (since these are technically overrideable).

The reason that we need to use EVAL is because to safely release a lock, we need to atomically check the lock identity before deleting it (to make sure it is not owned by some other process). Currently, the only way to do this in redis is through lau scripts (as far as I know).

I have actually been in a similar situation myself, and would like to see EVAL added to fakeredis or a similar library. However, for my case (running automated tests) it proved to be easy enough to create new redis containers on-demand, which ultimately provides more representative results.

from node-redlock.

Thank you for the detailed explanation. It does sound like setting up some test or on-demand Redis instances is the best option, and we’ll look into that.

from node-redlock.