In the context of chunked uploads we've observed that callers can legally provide content-encoding: aws-chunked, <something-else>as described in the docs
This extra metadata should become object metadata on S3, meaning S3 would use the aws-chunked value to understand the format of the incoming object, but store only the <something-else> part as metadata for the object.
Since the proxy uploads data to S3 in a single chunk (i.e., we never pass down aws-chunked data), we should make sure we still respect the content-type provided by the caller.
Also note that requests may have multiple content-encoding values and we should make sure the proxy can understand this regardless
In the context of chunked uploads we've observed that callers can legally provide
content-encoding: aws-chunked, <something-else>
as described in the docsThis extra metadata should become object metadata on S3, meaning S3 would use the
aws-chunked
value to understand the format of the incoming object, but store only the<something-else>
part as metadata for the object.Since the proxy uploads data to S3 in a single chunk (i.e., we never pass down
aws-chunked
data), we should make sure we still respect the content-type provided by the caller.Also note that requests may have multiple
content-encoding
values and we should make sure the proxy can understand this regardless