Improvement in fine grained file priorities (fgfp)

[ezalo]

Hi all

Nice new feature of yours, the fgpf. After using this feature, some ideas come to my mind. The ideas are the following

- set priority 0 for completed downloads

- recalculate priorities each time when the download of a file is finished

- only one file with high priority. I don't know if it is a bug but utorrent tries to download each high prioritized file (fine grained or not, as I noticed) with a nearly equal probability and, thus, this compromises the algorithm of prioritization.

Anyway, imho, fgpf must be always recalculated each time one file of the torrent is finished, since the original tail of the initially low prioritized files will remain for torrents with more than 15 files.

Of course, user can set 2 or more groups of files with fgfp. The algorithm above could be applied for all these groups, as they are distinguished by a sparse high priority file (or by discontinuities in the grained priorization). But it's not clear for me the trouble in the final implementation. Possibly, simply finding all the files until next discontinuity (or until the end of list) and then recalculating the priorities for each (now identified) group.

That's all. Thanks for your time and apologies if I messed with or missed something.

utorrent 2.2 Beta

[TDD_99]

Hi all

Nice new feature of yours, the fgpf. After using this feature, some ideas come to my mind. The ideas are the following

- set priority 0 for completed downloads

- recalculate priorities each time when the download of a file is finished

- only one file with high priority. I don't know if it is a bug but utorrent tries to download each high prioritized file (fine grained or not, as I noticed) with a nearly equal probability and, thus, this compromises the algorithm of prioritization.

Anyway, imho, fgpf must be always recalculated each time one file of the torrent is finished, since the original tail of the initially low prioritized files will remain for torrents with more than 15 files.

Of course, user can set 2 or more groups of files with fgfp. The algorithm above could be applied for all these groups, as they are distinguished by a sparse high priority file (or by discontinuities in the grained priorization). But it's not clear for me the trouble in the final implementation. Possibly, simply finding all the files until next discontinuity (or until the end of list) and then recalculating the priorities for each (now identified) group.

That's all. Thanks for your time and apologies if I messed with or missed something.

utorrent 2.2 Beta

First of all....... Thanks for the GREAT info!

2nd... I would never be able to explain this as well as you do here, but I totally understand you. I'm really glad to see ppl like you, who are able to explain things like this (that I do see/understand in my head, but never am able to explain myself)!

3th.... I know this post is kinda old, but it's the only post on regarding this "issue" and just needed to reply.

gr

Leroy