I am currently leaning toward implementing the http filesystem such that downloaded data is written to a temp file using the disk filesystem and read requests from from the public pdf_fsys_http_file_read function are handled with the pdf_fsys_disk_file_read function - being careful to only send back data that has actually been downloaded.
This will facilitate RIA, with some overhead of keeping track of what data has been written in the temp file. It will require setting the size of the temp file to the full size of the to-be-downloaded item right from the start, but this does not seem like a problem... Any comments would be appreciated. --Scott
