All committers have signed the CLA.

@emreyalvac, thank you. This is a very welcome addition, but there are some important points to consider:

1. It could be a breaking change for users who rely on the current replies.
2. The behavior should be similar for other write commands, not specific to JSON.SET.
3. Is it just about creating new keys, or also sub-paths for existing keys?

The best way to address 1. is probably to add a new configuration parameter (RedisJSON currently has none) that users can explicitly enable.

We also need a clear formal definition of the new behavior. JSONPath is tricky. Consider the following examples:

• JSON.SET j "$['a','b'].c" 1
• JSON.SET j "$[2:5:2].c" 1
• JSON.SET j "$..*.a 1
• JSON.SET j "$..[0]" [0]

@emreyalvac, thank you. This is a very welcome addition, but there are some important points to consider:

1. It could be a breaking change for users who rely on the current replies.

2. The behavior should be similar for other write commands, not specific to JSON.SET.

3. Is it just about creating new keys, or also sub-paths for existing keys?

The best way to address this is probably to add a new configuration parameter (RedisJSON currently has none) that users can explicitly enable.

We also need a clear formal definition of the new behavior. JSONPath is tricky. Consider the following examples:

• JSON.SET j "$['a','b'].c" 1

• JSON.SET j "$[2:5:2].c" 1

• JSON.SET j "$..*.a 1

• JSON.SET j "$..[0]" [0]

Thank you for your time. Working on it..