Hello, I have a question about the pk arguments.In my understanding, pk means the line of the table. If the developer create the model with the same pk, then the last one will be active and the old one will be covered. So, the developer will remember what's the current pk.
Currently, I use different global variables to save the pk for the different modules. It is not convenient and after program restart, the value will miss. So I will store it into redis later.

Do you have any idea about the correct way to pk ?
The following is the code.

`
devices_orm = DevicesModel(pk=1)
devices_orm_pk = 0
models_orm_pk = 0

def insert_redis(device):
global devices_orm
global devices_orm_pk
global models_orm_pk
devices_orm_pk = devices_orm_pk + 1
device_obj = DeviceModel(pk=devices_orm_pk)
device_obj.device_id = device["device_id"]

for model_object in device["models"]:
    models_orm_pk = 1 + models_orm_pk
    device_obj.models_list.lpush(
        MachineLearningModel(pk=models_orm_pk, name=model_object["name"], version=model_object["version"],
                             status=model_object["status"]))
devices_orm.devices_list.lpush(device_obj)

`

Object data from the local cache is inconsistent

for example

class SiteObject(models.Model):
    name = models.CharHash()

site1 = SiteObject(pk="site_pk_1")
site2 = SiteObject(pk="site_pk_1")
site1.name="pig"
print(site1.name)
site1.name = "cat"
print(site1.name)
print(site2.name)
site2.name="dog"
print(site2.name)
print(site1.name)

output

pig
cat
cat
dog
cat

As shown above, when site1.name is called, the data in redis is retrieved and stored in the site1 self._astra_hash object, so is site2, Therefore, the query of site1 will not be affected after the change of site2

Redis is so fast, I think the design might need to be rethought

I think this is a very good project and I hope more and more people will know about it.

How much work would it need to support 2.7?

Nice work @pilat !

Redis-astra will be shared in our next RedisWeekly tomorrow morning!