Closed stanorama closed 1 year ago
First, I'm not a Snowflake expert. I did see that approach, but as far as I could tell, it could also be problematic. My sense is that Snowflake should be used as a data warehouse, not in a web application context where concurrency is a factor. That being the case, adding more overhead each time a new object is created (perhaps even giving users the idea that this use case is appropriate) didn't seem worthwhile.
@sfc-gh-mkeller, any opinions?
Absent an argument that this is worth the cost, closing as wontfix.
I note in this project: https://github.com/benryan2010/django-snowflake-backend/blob/master/django-snowflake-backend/operations.py
They use the last_query_id to ensure that the max query relates only to the last query in the current session. I guess this only works if the database session is specific to the current web session and is not shared.
Would this help alleviate the race condition?