3 ms·
Wouldn't something like snapshot testing from a scheduled probe be more effective and reliable than using an LLM? Every X hours test the endpoints and validate
by cphoover 2y ago
Wouldn't something like snapshot testing from a scheduled probe be more effective and reliable than using an LLM?
Every X hours test the endpoints and validate the types and field names are consistent... If they change then trigger some kind of alerting mechanism to the user.
- alanloo 2y agoif the types and field names change, our parsing script should be able to detect that so it should be covered. I was talking about handling the subtle changes that are undetectable by checking field types and names