Charles Kerr
|
610c912c5a
|
(trunk libT) #3329 "connection problems when downloading" -- randomize the peer arrays so that all untested peers will have an equal chance of being used. Suggested by Longinus00
|
2010-06-22 16:31:38 +00:00 |
Charles Kerr
|
db44d81b24
|
(trunk) #3256 "libtransmission/publish.[ch] should be replaced" -- apply publish.diff for 2.10
|
2010-06-19 14:33:10 +00:00 |
Charles Kerr
|
e70bebf930
|
(trunk) #1521 "memory cache to reduce disk IO" -- commit block-cache-rc1.diff to trunk for the nightlies.
|
2010-06-19 14:25:11 +00:00 |
Charles Kerr
|
a5e708c824
|
(trunk libT) rejigger the filter order again, based on profiling in cachegrind
|
2010-06-14 12:48:28 +00:00 |
Charles Kerr
|
38a254ba78
|
(trunk libT) as long as I'm committing these CPU tweaks, might as well throw in #3289 too. To undo all this, we can revert to r10745
|
2010-06-14 12:01:50 +00:00 |
Charles Kerr
|
89201ab829
|
(trunk libT) when filtering out peers that aren't candidates, use the fastest filter tests first, so that the more expensive tests will be applied to fewer peers. In particular, call peerIsInUse() last.
|
2010-06-14 11:57:46 +00:00 |
Charles Kerr
|
f0a70d77b6
|
(trunk libT) experimental commit
|
2010-06-14 03:42:54 +00:00 |
Charles Kerr
|
ec79257999
|
(trunk libT) add a much faster (but more complicated) implementation of comparePeerCandidates(). For testing in the nightlies, leave both and the new implementations in the code and run them both (which will make the nightlies slower for awhile) and crash if the results differ from each other. When it doesn't crash anymore I'll remove the old implementation. Happy testing! >:)
-place and compare them to make sure they both give the same results. If no testers can make it crash
|
2010-06-14 01:56:03 +00:00 |
Charles Kerr
|
48f9d61ba1
|
(trunk libT) small speed improvements to comparePeerCandidates()
|
2010-06-13 23:06:50 +00:00 |
Mitchell Livingston
|
493f5afa6a
|
trivial improvement to compareSeedProbabilities()
|
2010-06-13 20:54:03 +00:00 |
Charles Kerr
|
a33be6c4b8
|
(trunk libT) speedups to comparePeerCandidates()
|
2010-06-13 17:17:33 +00:00 |
Charles Kerr
|
cf3417b2cd
|
(trunk libT) lower the max # of outbound peer connections a little
|
2010-06-04 16:48:36 +00:00 |
Charles Kerr
|
fcda98c86f
|
(trunk libT) promote PEX peers up the food chain a little bit
|
2010-05-19 15:53:09 +00:00 |
Charles Kerr
|
2b9ab542cb
|
(trunk) #3060 "Local Peer Discovery" -- in the code, rename LDS as LPD for Local Peer Discovery
|
2010-05-08 08:42:45 +00:00 |
Charles Kerr
|
5a34347f4f
|
(trunk) #3060 -- Local Peer Discovery patch from Eszet
|
2010-05-01 16:04:00 +00:00 |
Charles Kerr
|
97affa46a9
|
(trunk libT) #2609 "Leave 5% of connection slots for incoming connections" -- implemented in trunk for 2.00
|
2010-04-29 03:21:54 +00:00 |
Charles Kerr
|
9d3e02ce5c
|
(trunk libT) #2993 '"Downloaded" much greater than "Have" or "verified"' -- tweak the download throttle algorithm a bit to try & address the slowness reported by Rolcol @ http://trac.transmissionbt.com/ticket/2993#comment:42
|
2010-04-22 03:12:31 +00:00 |
Charles Kerr
|
546bca723b
|
(trunk libT) #3159 "better decision-making when choosing which peers to initiate new connections with" -- experimental commit
|
2010-04-20 21:54:03 +00:00 |
Charles Kerr
|
b7827565ee
|
(trunk libT) #1869 "new status for torrents that reach seed ratio" -- possible fix for the bug reported by leena
|
2010-04-15 13:43:54 +00:00 |
Charles Kerr
|
0193a577b9
|
(trunk libT) increase request TTL to 120 seconds
|
2010-03-22 02:27:57 +00:00 |
Charles Kerr
|
fe3d9ed95a
|
(trunk) one of the least interesting commits ever: remove trailing spaces from source code lines...
|
2010-03-17 17:07:40 +00:00 |
Charles Kerr
|
1d8d9d8447
|
(trunk libT) code cleanup: after r10346, we don't need to keep the per-torrent cancel/block histories anymore
|
2010-03-10 15:55:00 +00:00 |
Charles Kerr
|
3aa3ec7d14
|
(trunk libT) #2993 "'Downloaded' much greater than 'Have' or 'Verified'" -- tweak the new throttle code based on slow download feedback from gn0s1s in irc, from Rascal @ http://trac.transmissionbt.com/ticket/2993#comment:29, and from AGSystem @ http://forum.transmissionbt.com/viewtopic.php?p=45631&f=1#p45631
|
2010-03-10 15:34:27 +00:00 |
Charles Kerr
|
b850349479
|
(trunk) #2993 "'Downloaded' much greater than 'Have' or 'verified'" -- add new congestion-based throttle for 2.0
|
2010-03-08 04:29:58 +00:00 |
Charles Kerr
|
b1a29969fe
|
(trunk libT) disable the comparePieceByWeight assertions in trunk too.
|
2010-03-07 22:56:14 +00:00 |
Charles Kerr
|
095cc3e05f
|
(trunk libT) "don't cancel requests for blocks that we're downloading from slow peers" -- fixed in trunk for 1.92
|
2010-03-06 14:56:15 +00:00 |
Charles Kerr
|
2be28fe9dd
|
(trunk libT) #2993 "'Downloaded' much greater than 'Have' or 'Verified'" -- found a weighted piece sorting issue while trying to confirm or refute this issue.
|
2010-03-03 04:16:18 +00:00 |
Charles Kerr
|
f1221f958f
|
(trunk libT) dead code removal -- the buffer & buflen arguments to tr_ioTestPiece() were unused
|
2010-02-23 06:05:18 +00:00 |
Charles Kerr
|
1499f5c405
|
(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92(trunk libT) #2906 "remove unused code: Torrent.requests array is never sorted by time" -- added to trunk for 1.92
|
2010-02-22 15:07:14 +00:00 |
Charles Kerr
|
a973e38bb5
|
(trunk libT) #2922 "announce's 'download' field should follow the de facto standard" -- added to trunk for 1.91
|
2010-02-20 15:36:25 +00:00 |
Charles Kerr
|
1724553212
|
(trunk) per-tier up/down/corrupt byte counts for announcing to trackers
|
2010-02-08 16:47:30 +00:00 |
Charles Kerr
|
9489762d26
|
(trunk libT) remove the last two experimental commits... nightly users, it's probably safe to use svn again. or at least as safe as usual... ;)
|
2010-02-03 01:00:08 +00:00 |
Charles Kerr
|
825d5c3a5a
|
(trunk libT) this is an experimental commit for a debugging session with people who don't have access to build tools. unless you're one of those testers, or unless you enjoy not connecting to peers, you DO NOT WANT TO USE THIS REVISION
|
2010-02-03 00:39:27 +00:00 |
Charles Kerr
|
bdb706edb9
|
(trunk libT) this is an experimental commit for a debugging session with people who don't have access to build tools. unless you're one of those testers, you might not want to use this revision.
|
2010-02-03 00:28:07 +00:00 |
Charles Kerr
|
30f85172c4
|
(trunk libT) taper off the number of new connections per pulse per torrent based on how long the torrent's been running. Brand-new torrents get a higher burst of new peer connection attempts, but long-running torrents don't need that kind of activity.
|
2010-02-03 00:12:19 +00:00 |
Charles Kerr
|
f010687d81
|
(trunk libT) #2792 "Since 1.80 update downloads failing" -- a better fix than r10015...
|
2010-01-25 05:19:54 +00:00 |
Charles Kerr
|
bfbc126ea0
|
(trunk libT) #2792 "Since 1.80 update downloads failing" -- fixed in trunk for 1.83
|
2010-01-25 04:54:04 +00:00 |
Charles Kerr
|
8e16b74dad
|
(trunk libT) #2775 "Saving some memory/storage" -- committed for 1.80
|
2010-01-20 18:48:52 +00:00 |
Charles Kerr
|
9eb07e9b16
|
(trunk) No code changes here... filling in some of the blanks in the "peers" and "utils" doxygen groups' documentation.
|
2010-01-19 19:37:00 +00:00 |
Charles Kerr
|
4b0c1ea332
|
(trunk libT) #2705 "speeding up request queue management" -- committed v1.1.diff for 1.80
|
2010-01-12 20:17:22 +00:00 |
Charles Kerr
|
182cbb8b90
|
(trunk) one of the less-interesting commits in a while: remove trailing spaces from lines
|
2010-01-05 23:47:50 +00:00 |
Charles Kerr
|
5c26afdb52
|
happy new year!
|
2010-01-04 21:00:47 +00:00 |
Charles Kerr
|
6f68e30f37
|
(trunk libT) consistency tweak: rename tr_global{Lock,Unlock,IsLocked} as tr_session{Lock,Unlock,IsLocked}
|
2010-01-04 09:05:02 +00:00 |
Charles Kerr
|
6b8fb48d64
|
(trunk libT) replace the TR_INLINE macro with the standard "inline" keyword, since we require inline functions anyway
|
2010-01-01 22:26:35 +00:00 |
Charles Kerr
|
1eb0cdd7c6
|
(trunk libT) fix various minor compiler warnings that show up when you build libtransmission with NDEBUG defined
|
2009-12-28 23:27:17 +00:00 |
Charles Kerr
|
3e2269e5e3
|
(trunk libT) #2632 "Add streaming capability to libtransmission (but not the Transmission GUI clients)" -- revert this feature due to pushback from devs and users and the 1.80 freeze
|
2009-12-24 18:01:48 +00:00 |
Charles Kerr
|
12ffdb7d87
|
(trunk libT) #2632 "Add streaming capability to libtransmission (but not the Transmission GUI clients)" -- implemented
|
2009-12-24 01:02:54 +00:00 |
Charles Kerr
|
d9369718b3
|
(trunk) #2548 "T's request queue can send out too many duplicate requests" -- (1) fix r9465 implementation bug that caused some peers to get starved of requests if they rejected a request or choked, then unchoked us. (2) increase a block request's TTL by 15 seconds to reduce cancel/req cycles between two or more blocks (3) add a debug mode to the GTK+ client's peer tab to watch the pending requests counts both up & down
|
2009-12-16 18:20:01 +00:00 |
Charles Kerr
|
f3e8a29e0f
|
(trunk libT) #2548 "T's request queue can send out too many duplicate requests" -- fix off-by-one when counting how many peers we'll send a given block request to during endgame
|
2009-12-15 21:33:24 +00:00 |
Charles Kerr
|
22e33f784b
|
(trunk libT) #2548 "T's request queue can send out too many duplicate requests" -- experimental revision to r9465 to address further overdownloading issues
|
2009-12-15 20:13:34 +00:00 |