We commonly use the OAI PMH protocol to allow customers to harvest our data. OAI PMH fits well for some of our data, but not for others. Ie, for some data it is a forced fit. So I am looking for other "standards" or "common practices." Aspects of OAI PMH I like are
It is record oriented
There are sets of records and records can be in multiple sets
Each record has an identifier, a relevant date, a list of sets, and content
The identifiers can be harvested that match optional criteria of set and date period
The content can be harvested that match optional criteria of set and date period
All harvesting is an iterative process of getting an unordered collection of records until no records remain.
It is not necessary to know ahead of time the count of records harvested (by either the server or the client).
Can anyone recommend an existing HTTP-based protocol that can span content from TSVs data to ZIPs of files?