f87b732959
added the initial version of the (t *Torrent) KnownSwarm() function
2017-09-12 15:22:53 +01:00
bad6f07f5e
Wrap writerCond in tickleWriter and be more selective about using it
2017-09-01 15:26:50 +10:00
13e79039f2
Generate requests, cancels and interest state in the connection writer
2017-08-31 23:48:52 +10:00
0388ec1eee
Synchronize Torrent.Info
2017-08-29 15:16:53 +10:00
a8a1ea755d
Remove unused functions
2017-08-26 13:23:04 +10:00
76c60ffa77
Try a state-delta function for updating request state
...
Also adds Torrent.networkingEnabled, though it isn't yet useful.
2017-08-18 01:51:02 +10:00
493916c279
Rename Torrent.connHasWantedPieces->connection.peerHasWantedPieces
2017-08-18 01:48:19 +10:00
824441efa0
Move logging inside client lock, for Torrent.String
...
Fixes #163
2017-07-01 16:01:38 +10:00
a3d231cb92
signal an event when torrent closed
2017-06-05 08:46:50 +08:00
296ef76c36
Tidy up the duplicate conn selection code
2017-05-26 13:00:28 +10:00
5bde8de95d
perf.Timer.Stop renamed
2017-05-26 13:00:28 +10:00
d8023a5e39
bug fix: race condition between outgoing connection and incoming connection
2017-05-10 18:31:19 +08:00
0d19c72ea5
Added peer count stats to TorrentStats struct
2017-04-22 23:29:10 -04:00
ba3e798b5f
Fix a logging of metainfo.Hash
2017-02-24 21:56:23 +11:00
aa74d992d3
Don’t track the empty string, and ditch Torrent.announceList
2017-02-19 15:57:30 +11:00
350dd01ab9
Fix torrent status info hash hex
2017-02-16 20:10:32 +11:00
a959bdd0f8
Revert "Quick fix for missing MetaInfo.Announce everywhere with trackers"
...
This reverts commit b7a8d08bbf
2017-02-15 18:40:30 +11:00
b7a8d08bbf
Quick fix for missing MetaInfo.Announce everywhere with trackers
2017-02-13 22:28:42 +11:00
b868a6b9e2
dht.Server.Announce now takes [20]byte
2017-02-07 15:01:02 +11:00
79d800fe92
Trivial logging, comment, Stringer changes
2017-02-02 16:53:19 +11:00
55b6025f80
Log the storage TorrentImpl type
2017-01-04 18:15:49 +11:00
9cf09dba8a
Remove unused funcs and idents
2017-01-01 11:03:02 +11:00
186b5073d2
Move method queuePieceCheck
2017-01-01 11:02:37 +11:00
d48c7c8e36
sort imports
2017-01-01 11:01:41 +11:00
b68d7cd08e
dht/… moved to github.com/anacrolix/dht
2016-12-14 11:43:37 +11:00
f510c5cf22
Fix piece failures not clearing dirty chunks
...
I don’t like how complicated piece completion/hashing management is, but it works.
2016-12-06 16:10:10 +11:00
af7a0d3464
Add Torrent.Closed
2016-11-30 18:02:39 +11:00
1f2f037209
Fix logging of trust slice
2016-11-30 18:01:39 +11:00
f451a39b69
Fix harmless race conditions in Client.WriteStatus introduced way back
2016-11-27 14:26:45 +11:00
4a7fbf6170
Add peers received from received announce_peer DHT messages to the Client
...
Addresses #133
2016-11-27 00:05:19 +11:00
fdf75d59c8
Kick only the worst connection when a piece fails a check
2016-11-23 12:59:23 +11:00
57b679ffb0
Add a note about dropping connections that we sent a HAVE for incomplete pieces
2016-11-23 11:52:41 +11:00
4807c9e29a
Make Torrent.conns a map
...
Can't remember if I've tried this before. But dropping arbitrary connections while iterating established conns will become much simpler.
2016-11-23 11:48:44 +11:00
ebbd555e7b
Move a bunch of Client methods onto more appropriate types
2016-11-22 21:12:53 +11:00
90ca45dd2d
Don't accept pieces with bad indexes
...
Fixes #132
2016-11-22 14:20:48 +11:00
df60a765a7
Torrent.unpendPieces: Only update unpended pieces
...
Updating all pieces at once is no longer faster. (There was an optimization for this a while ago that no longer exists.)
2016-10-31 19:05:08 +11:00
a8e96ce996
When Reader position changes only update piece priorities that may be affected
...
Seems to significantly reduce CPU load reading from torrents with huge piece counts.
2016-10-31 19:00:08 +11:00
95f73db7e4
Ditch the use of generic slices.HeapInterface for finding the worst "bad" connection
...
It's become a bottleneck. Implement it directly using heap.Interface.
2016-10-31 16:24:48 +11:00
1725133111
Move the implementation of connection.requestPiecePendingChunks out of Torrent
2016-10-25 14:58:17 +11:00
0e221dbdcd
Optimize piece priorities when reader position changes
...
Gives a decent boost to throughput and reduces a lot of CPU when reading very quickly from Reader.
2016-10-23 16:33:26 +11:00
de761fb506
Do chunk pooling at Torrent instead of connection level
2016-10-05 15:57:00 +11:00
1e919dd6b1
Rework storage interfaces to make them simpler to implement
...
This allows lots of behaviour to be baked into the new Client, Torrent and Piece wrappers, rather than duplicating (badly) them in all the backend implementations.
2016-09-02 15:10:57 +10:00
8b17d2a63a
Rework Reader position changes affecting piece priorities
...
Torrent.updatePiecePriorities is always followed by Torrent.piecePriorityChanged, so move it into the former function. Also drop the separate mutex for Reader, and add a missing Reader.posChanged call.
2016-08-30 15:41:26 +10:00
6c15a030d6
Improve comments
2016-08-30 15:07:59 +10:00
2a1cef7c9e
Remove the InfoEx type, and don't generate its infohash on the fly
...
Fixes #106 .
2016-08-26 20:29:05 +10:00
6065a55818
Fix some more races in the status writing
2016-08-02 11:18:33 +10:00
1ca2e7e95a
Add some extra conditions to prevent network activity when a torrent is closed
...
From suggestions by @axet, this might be a thing sometimes.
2016-07-30 00:41:45 +10:00
4143b7470d
Begin exposing all the magic that Client.WriteStatus has so people can parse Client state their own way
...
It wasn't possible to move Client.WriteStatus to an external package to verify it doesn't depend on same-package access to data because then it can't be used in the tests, and it's extremely useful there. So I've settled for not locking the Client, and trying to use all the public methods. It's a work in progress.
2016-07-30 00:37:52 +10:00
49315045a0
it is actually sha1 hash size
2016-07-29 22:50:58 +10:00
3c75f68413
Kick connections that contribute the final piece of missing metadata that fails to validate
...
Failing to get metainfo was resetting Torrent.metadataBytes, which caused arriving metadata data to appear to belong to nonexistent pieces, despite passing the check that we'd in fact requested them from the sending connection. This was unnecessarily noisy. Instead return an error if we got all the data, but couldn't set the Torrent's info bytes, that should propagate out and cause the connection that contributed the final missing piece to be dropped. This will at least provide some improved resistance to unfriendly behaviour on the network.
2016-07-24 00:34:40 +10:00
Bora M. Alper