We have a use case to expose Fedora datastream/objects to science data tools that are dependent on a meaningful file system and MIME indicators by file extension. For example, Earth and Atmospheric Science community has a tool to read a special format called NetCDF that represents multiple-dimensional arrays and there are data tools serve them via HTTP, but needs to scan the file system for that type of files, rather than read a HTTP stream.
I would therefore like to have the the hints argument in the Akubra API to be populated with the mime type and the parent PID, in order to implement a special storage on the file system for this kind of tool. The Parent PID is to be used to form a correct collection hierarchy, and the MIME type is to be used to write the file extension.
We have had an implementation (using the Fedora 3.4.2 legacy storage system) that used the internal low-level storage API, which needs to be move into Akubra as a true sustainable solution.