I’m curious - is there a reason Snowplow collectors don’t validate JSON events at the point of collection and return a bad response for rows that don’t match their schema? I have faced a few situations recently where this would have saved a huge amount of head-scratching.
This it obviously inherently impossible for the CloudFront collector but it seems to me straightforward for the others and potentially very useful. I can see a few cases where some decisions would need to be made (e.g. what happens to a request with multiple events) but Is there a philosophical reason why it hasn’t been implemented?