Hello, ?J?r?me! On Sat, 2008-02-23 at 16:12 +0100, J?r?me Chabod wrote: > It works fine. Ok, I'll apply it, then :) > However, I notice pourcentage calculation is not accurate enougth when > episodes size is very different, because it is base on the average > pecentage of downloading episodes. It should rather be based on ratio > total volume done/ total volume remaining (it would also fix the > problem with cancelled episode). For that, we have to know the file size of each episode before the download. This works for most podcast feeds, but doesn't work for some others. The solution to this problem would be: http://bugs.gpodder.org/show_bug.cgi?id=24 which depends on the SQLite-based storage method: http://bugs.gpodder.org/show_bug.cgi?id=12 After we have that, we should be able to determine the file size of every episode regardless of the contents of the RSS feed's "length" attribute of the "enclosure" element. We should then be able to use the file size as a way of determining the "bytes done" and "bytes todo" values of our downloads. > Have a try: allow only one episode in the queue, and start dowloading > a small episode then a big one. The percentage will go very quick from > 0 to 50% then very slow from 50 to 100%. The effect on progress bar is > not very nice, and the estimated remaining time completly incorrect > > I you agree, I can try to fix it. I agree ;) I think it would be best to file a new bug and let it depend on bug #24. Of course, that's just my proposal, so if you have any other idea how this could be implemented so that it works even for queued (i.e. not yet started) downloads and for RSS feeds with missing or invalid "length" field, that's fine, too :) Thanks, Thomas