The ActionDownload implementation usually requires an inputstream to be returned. Currently we implement a FileInputStream from the document to be view physically being on disk.
The currently works fine.
We are looking at removing the need to output files from the DMS to disk as this requires managing our own cache and if we can remove the need that would be great.
The API for the DMS we use give the option to return you an OutputStream object, from what i can tell i can convert the output to an input stream but i am concerned about memory use as the entire document will be held in memory completely until the it has been written out to disk.
I have been reading up on PipedOutStream and PipedInputStreams and if they would be beneficial in my use case scenario?