what gets me mad is when some NMDC client has thrown some error and it downloads the same file for eturnity. As soon as the client finishes it starts again. I would like to give that slot to a more deserving user.
could a feature be added to detect that sort of behavior from a client and just not give them a slot?
nmdc FIX in dc++
Moderator: Moderators
-
GargoyleMT
- DC++ Contributor
- Posts: 3212
- Joined: 2003-01-07 21:46
- Location: .pa.us
How about just if they have redownloaded a certain number of times (maybe a threshold of no more than 3?) then cut them off. It does kind of make sense as if they haven't gotten it by then, then they may very well never get it sucessfully and it's rather a pointless waste of bandwidth considering that they should have already had the file after all that time.
The only thing is that it would be hard to track that. Either it would have to keep some kind of list sorted by time where it removed data after it's been there so long (otherwise the list will become absolutely gigantic and slow the entire thing down horribly with processing) or it would have to just hold things in memory in which case it would have to have a similar limitation, but it could afford to get a little larger. It also occured to me that it could have different limit methods like max filesize/etc, but I think that remembering with a time system is best if such a thing were to be done.
BTW, I'm not entirely innocent of this myself. Occasionally a file gets screwed up on mine and I'm forced to redownload it because it just can't download to the file anymore no matter how much rolling back. That's why I definitely think there needs to be a threshold point on that.
The only thing is that it would be hard to track that. Either it would have to keep some kind of list sorted by time where it removed data after it's been there so long (otherwise the list will become absolutely gigantic and slow the entire thing down horribly with processing) or it would have to just hold things in memory in which case it would have to have a similar limitation, but it could afford to get a little larger. It also occured to me that it could have different limit methods like max filesize/etc, but I think that remembering with a time system is best if such a thing were to be done.
BTW, I'm not entirely innocent of this myself. Occasionally a file gets screwed up on mine and I'm forced to redownload it because it just can't download to the file anymore no matter how much rolling back. That's why I definitely think there needs to be a threshold point on that.