Current setup:

http://URL/pieceCID returns a unpadded car file of the data

http://URL/pieceCID.car returns a padded car file of the data (whole piece)

http://URL/payloadCID returns a unpadded car file of the data

http://URL/payloadCID.car returns a padded car file of the data (whole piece)

The distinction between the URLs is to allow users to download either by specifying the pieceCID or payloadCID whatever is available to them.

But this gets confusing because payload makes you think that we are talking about the unixfs data here.

Perhaps, we can do it like:

http://URL/pieceCID returns a unpadded car file of the data

http://URL/pieceCID.car returns a padded car file of the data (whole piece)

http://URL/payloadCID returns a unixfs data

http://URL/payloadCID.car returns a unpadded car file of the data

or rename "payload" to something else to confusion between data and car file. "Payload" also gives you incorrect idea that you can retrieve partial DAGs from within the uploaded car files