#4931 closed Bug (fixed)
Wrong peer percentages for magnetized transfers
Reported by: | leandroong | Owned by: | jordan |
---|---|---|---|
Priority: | Normal | Milestone: | 2.60 |
Component: | libtransmission | Version: | 2.52 |
Severity: | Normal | Keywords: | patch needed |
Cc: |
Description
See this image [site:]http://www.flickr.com/photos/66237389@N07/7312212828/
Change History (10)
comment:1 Changed 11 years ago by cfpp2p
- Keywords patch needed added
comment:2 Changed 11 years ago by leandroong
It is enough for me that you know that problem exist and need fixing in the future. Minor bug only.
comment:3 Changed 11 years ago by livings124
- Summary changed from Displayed over 20400% on one of the peers to Wrong peer percentages for magnetized transfers
comment:4 Changed 11 years ago by jordan
looks like this is coming from libtransmission's tr_peerUpdateProgress() code... there's even a comment without pieceCount, this result is only a best guess... which is maybe too generous as "best" isn't very good here. :)
comment:5 Changed 11 years ago by jordan
- Component changed from Transmission to libtransmission
- Milestone changed from None Set to 2.60
- Owner set to jordan
- Status changed from new to assigned
comment:6 Changed 11 years ago by jordan
- Resolution set to fixed
- Status changed from assigned to closed
r13360 clamps that guess in the range of [0.0...1.0] :)
comment:7 Changed 11 years ago by cfpp2p
but the clamp is still just a guess so sometimes we'll see say 0%, then immediately jump to 100%. The fix just appeases the user so it looks nicer to see 0% than 20400% but in the end both are still very inaccurate guesses.
comment:8 Changed 11 years ago by leandroong
Version: 2.60+_13375 Good enought for me. Thanks
comment:9 Changed 11 years ago by jordan
cfpp2p: I don't think we can have an accurate guess until we know how many blocks/pieces there are in the torrent...
comment:10 Changed 11 years ago by x190
Therefore, display N/A or leave it blank. That way users won't be left wondering why magnet links "won't work" when they have all these 100%ers who may actually only have 100 bytes.
this as far as i can tell only happens when still acquiring the metadata for magnet links (therfore the total size for the torrent is still unknown) and then the % the peer has is miscalculated since we don't know the actual size of the torrent until after the metadata is aquired. You could try patching so that tr_torrentHasMetadata() is checked and display a '?' or something like that until tr_torrentHasMetadata() returns true. Just an idea, but haven't really looked into it.