Closed mcwhittemore closed 7 years ago
Actually it never should've rejected anything -- the user was able to set a limit on the maximum size of a feature that will be stored in DynamoDB, and larger than that feature data was stored on S3.
@rclark - yea, but now that we don't use s3, will this make it possible to write items to DynamoDB that you can't get out?
If you call cardboard.put
on a feature that won't fit into dynamo, you'll get a DynamoDB error response handed back to you. So maybe the question is whether you want to just pass that back out or make cardboard's error types a part of its API contract.
Oh. Duh. DynamoDB won't let you write features that are too big. I think AWS errors are ok for Cardboard. If someone doesn't want them, they should validate feature size on their own end.
This is something that got dropped with the move to Cardboard@3.0.0. Should it come back?