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
|
22e61e68ed
|
(trunk libT) fix call to updateDesiredRequestCount() reported by sadface
|
2009-12-16 01:32:23 +00:00 |
|
Charles Kerr
|
d6f7df8c9a
|
(trunk libT) #2508 "atom->port never updated" -- (1) fix r9434 bug that used the wrong byte ordering scheme for port numbers. This is why most of the .resume file's peers are worthless in betas 1 and 2. :( (2) in the resume file, change the dictionary key for peers so that the worthless peers won't be reloaded on startup in beta 3.
|
2009-12-16 00:45:18 +00:00 |
|
Charles Kerr
|
c819d6dca6
|
(trunk libT) #2671 "downloading from too many peers" -- address a smaller aspect of this ticket, by reducing the minimum number of blocks we request from a peer, as noted by jch's comment @ http://trac.transmissionbt.com/ticket/2671#comment:2
|
2009-12-15 18:11:20 +00:00 |
|
Charles Kerr
|
f4c744266f
|
(trunk libT) #2667 "ignore martian addresses" -- modified this ticket. previously we ignored them from LTEP; now we ignore them regardless of the source. this commit implements this filtering change.
|
2009-12-15 17:39:19 +00:00 |
|
Charles Kerr
|
3fe5db7113
|
(trunk libT) #2667 "Ignore martian addresses in LTEP messages" -- implemented for 1.80
|
2009-12-14 02:07:47 +00:00 |
|
Charles Kerr
|
907c67ba78
|
(trunk libT) remove unnecessary #includes
|
2009-12-07 03:57:55 +00:00 |
|
Charles Kerr
|
d370222113
|
(trunk libT) #2551 "when uploading to peers, prefetch local data from disk" -- prefetching performance tweak described in http://trac.transmissionbt.com/ticket/2551#comment:22 -- instead of prefetching [8..12] blocks, always prefetch 12 blocks
|
2009-12-05 19:30:15 +00:00 |
|
Charles Kerr
|
78ead8c3dd
|
(trunk) update the copyright notices
|
2009-12-05 02:19:24 +00:00 |
|
Charles Kerr
|
7537f02589
|
(trunk libT) #2636: "wrong peer progress update" -- possible fix as described in comment:2
|
2009-12-03 03:11:22 +00:00 |
|
Charles Kerr
|
47c6ec5efa
|
(trunk libT) #2551: "when uploading to peers, prefetch local data from disk" -- fix peer-msgs.c NULL pointer dereference introduced in r9514 for this ticket. Reported by Waldorf
|
2009-11-27 03:21:21 +00:00 |
|
Charles Kerr
|
e1c6b792aa
|
(trunk libT) #2610 "avoid unnecessary calls to time(NULL)"
|
2009-11-26 18:47:08 +00:00 |
|
Charles Kerr
|
1101f6710e
|
(trunk libT) #2430 "Peer atom pool grows too large" -- add an atom expiration system along the lines of the suggestions in this ticket's comments. jch and KyleK, let me know if you think anything in this commit needs improvement.
|
2009-11-26 05:13:58 +00:00 |
|
Charles Kerr
|
c102f199c3
|
(trunk libT) don't allow duplicate torrents to be added via magnet links
|
2009-11-24 17:31:20 +00:00 |
|
Charles Kerr
|
d5ae900e4e
|
(trunk) #2096 "magnet links": in libT, cleanup. in GTK+ client, add different status text in the main window for magnet torrents: "downloading .torrent data"
|
2009-11-24 15:51:16 +00:00 |
|
Charles Kerr
|
f69e335291
|
(trunk) #2096: Magnet, BEP #9 support
|
2009-11-24 02:16:31 +00:00 |
|
Charles Kerr
|
1cbbcf9fa1
|
(trunk libT) jch's patches 0001 through 0004 for ticket #2576, IPv6 support for DHT (BEP #32)
|
2009-11-24 01:59:51 +00:00 |
|
Mitchell Livingston
|
be85dc0fac
|
a couple of more swarm speed removals
|
2009-11-22 01:12:44 +00:00 |
|
Charles Kerr
|
ec08e333a2
|
(trunk libT) fix a stupid trunk-only bug that messed up which peer requests we responded to. maybe this has something to do with the slower speeds Rolcol's seeing?
|
2009-11-22 00:52:11 +00:00 |
|
Charles Kerr
|
22f7a64aa2
|
(trunk libT) #2575: PEX is not meant to do forwarding
|
2009-11-20 07:47:31 +00:00 |
|
Charles Kerr
|
cf66227f0d
|
(trunk libT) #2573: Don't send BT_PORT messages over IPv6
|
2009-11-12 01:59:53 +00:00 |
|
Charles Kerr
|
1f0fdba2e8
|
(trunk libT) #2551: when uploading to peers, prefetch local data from disk
|
2009-11-10 17:26:04 +00:00 |
|
Charles Kerr
|
44df1d5548
|
(trunk libT) #2112: provide "ipv6=" parameter to trackers
|
2009-11-10 17:03:23 +00:00 |
|
Charles Kerr
|
03e9189333
|
(trunk libT) #2572: don't send empty dictionaries when PEXing
|
2009-11-10 16:32:14 +00:00 |
|
Charles Kerr
|
225384fcc5
|
(trunk libT) #2502: announce own IPv6 address to peers
|
2009-11-09 05:45:16 +00:00 |
|
Charles Kerr
|
b906c125ac
|
(trunk libT) #2548: T's request queue can send out too many duplicate requests
|
2009-11-08 23:20:00 +00:00 |
|
Charles Kerr
|
4d817b0701
|
(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...(trunk libT) undo r9465 + r9466, which was the experimental new request manager. It still still needs some tinkering before it's ready for the nightlies...
|
2009-11-02 00:17:30 +00:00 |
|
Charles Kerr
|
296411e093
|
(trunk libT) turn off a debugging message in the terminal
|
2009-11-01 03:56:10 +00:00 |
|
Charles Kerr
|
0a4d8602e0
|
(trunk libT) #2548: T's request queue can send out too many duplicate requests
|
2009-11-01 02:10:47 +00:00 |
|
Charles Kerr
|
3680896a19
|
(trunk libT) #2508 atom->port never updated
|
2009-10-29 16:10:03 +00:00 |
|
Charles Kerr
|
00197cc461
|
(trunk libT) undo today's earlier commit r9381 for #2508 -- looks like I introduced a bug into the patch.
|
2009-10-22 19:08:45 +00:00 |
|
Charles Kerr
|
bb541fc8d6
|
(trunk libT) #2508: atom->port never updated
|
2009-10-22 15:05:56 +00:00 |
|
Charles Kerr
|
aad2a886be
|
fix oops
|
2009-10-03 13:17:18 +00:00 |
|
Charles Kerr
|
412f0b40a2
|
(trunk libT) fix a crash reported by John Clay and silence a gcc warning
|
2009-10-03 13:15:32 +00:00 |
|
Charles Kerr
|
6656c6d71e
|
(trunk libT) #2420 - Transmission should follow PEX conventions
|
2009-09-17 19:05:18 +00:00 |
|
Charles Kerr
|
034d14639a
|
(trunk libT) use tr_torrentName() instead of tor->info.name
|
2009-09-17 03:24:35 +00:00 |
|
Charles Kerr
|
fb83a0464e
|
(trunk libT) #2395: caching too many peers between sessions makes startup slow and uses too much memory (KyleK, stdisease, et al)
|
2009-09-10 02:20:35 +00:00 |
|
Charles Kerr
|
6dedb0433b
|
(trunk libT) better throttling back of outgoing download requests when speed limits are turned on
|
2009-08-18 01:57:17 +00:00 |
|
Charles Kerr
|
11717de1d0
|
(trunk libT) fix #2332: filesystem-based error messages lack context
|
2009-08-13 17:25:26 +00:00 |
|
Charles Kerr
|
cb79214854
|
(trunk) remove trailing spaces
|
2009-08-10 20:04:08 +00:00 |
|
Charles Kerr
|
24746cff7f
|
(trunk libT) minor #include futzing
|
2009-07-01 14:58:57 +00:00 |
|
Charles Kerr
|
1980bcc988
|
(trunk libT) I love you, massif pt. 2
|
2009-06-25 17:25:51 +00:00 |
|
Charles Kerr
|
8d52cc2adb
|
(trunk libT) scrimp & save a few more bytes per each peer
|
2009-06-24 04:14:39 +00:00 |
|
Charles Kerr
|
1be7f51b3d
|
(trunk libT) simplify the evtimer / struct timeval code a little
|
2009-06-15 03:24:40 +00:00 |
|
Charles Kerr
|
0380906aa6
|
(trunk libT) a minor but nice twofer: reduce per-peer memory overhead a little bit, and also avoid an extra malloc/free per peer connection
|
2009-06-15 02:22:41 +00:00 |
|
Charles Kerr
|
813640d6f3
|
(trunk libT) silence a gcc 4.4 warning: "dereferencing type-punned pointer will break strict-aliasing rules"
|
2009-06-14 19:17:10 +00:00 |
|
Charles Kerr
|
ef21704b58
|
(trunk libT) remove the last calls to tr_getBuffer() and tr_releaseBuffer()
|
2009-06-14 01:00:36 +00:00 |
|
Charles Kerr
|
0dd0a0bbd7
|
(trunk libT) #2195: 1.71 assertion: tr_peerIoSupportsFEXT( msgs->peer->io ) peer-msgs.c:265
|
2009-06-10 13:46:46 +00:00 |
|
Charles Kerr
|
5898c8b276
|
(trunk libT) add a debugging message to r8609 in case we ever have to go back to the bad old days of TR_DEBUG_FD
|
2009-06-04 15:04:34 +00:00 |
|
Charles Kerr
|
cc4cd19903
|
(trunk libT) under very rare circumstances we wouldn't send a REJ message even if the protocol required it. also, avoid an unnecessary memcpy(). Thanks to geirha and Lacrocivious for helping test & confirm the fix.
|
2009-06-04 14:59:17 +00:00 |
|