> > hm, so the uuid seems to be there to take into account, that multiple > files with the same filename might be uploaded. That makes sense, > exspecially if not all files README.txt have the same content. >
If the filename is very long, part of it might get cut off depending on the "length" attribute of the upload field (particularly since the b16encode increases the name length), so even files with different original names could end up with the same truncated b16encode value. The uuid will also prevent someone from guessing the URL for a file even if they know (or correctly guess) the original filename. > Interesting enough I was able to totally skip the file database table > completely. > I got a unique name for my data sets and _generate_ files with unique > filenames from those datasets. > I know there is a dataset P03-1232012312. > So if I want to download the file P03-1232012312_EN.pdf I need to call > .../download/t_files.f_file.xxx.5030332d31323332303132333132.pdf > Are you using the standard download function, which calls response.download()? If so, that only works by checking the db. Or having you created a custom download function that just opens the file and streams it? Anthony
