0827f5e955
gx publish 0.9.31
2018-10-04 19:23:53 -04:00
1abb0209ce
gx publish 0.9.30
2018-10-04 18:41:48 -04:00
8680a79d3a
add Join/Leave to PubSubRouter interface
...
advanced routers will want to know when they are subscribed to a topic as
they may want to send control messages.
2018-06-12 08:00:17 -07:00
75fcbf0c2c
New -> NewFloodsubWithProtocols
2018-06-07 09:21:08 -07:00
6840b190db
allow FloodSubRouter to be configured with protocols
2018-06-07 09:21:08 -07:00
327e72d4fd
demote erros per #62
2018-01-27 09:54:57 +02:00
e7b1fe6e75
modularized pubsub; Flooding is just a routing method.
2018-01-27 09:54:57 +02:00
098f1d5819
demote errors/warnings to info
...
* dropped/killed connections are not errors.
* handleIncomingRPC/publishMessage do not return errors.
2018-01-26 15:00:47 -08:00
c82e67dcd3
Merge pull request #55 from libp2p/feat/validators
...
Message Validators
2018-01-24 06:03:20 +00:00
3f4fc21228
fix comment, subscriptions don't have validators any more.
2018-01-18 20:09:09 +02:00
bbdec3fda2
implement per topic validators
2018-01-18 19:12:36 +02:00
99a4f9e5ed
Increase timeout in TestSubscribeMultipleTimes
...
Increase delay to 10ms; makes go test -race happy.
2018-01-16 11:46:33 +02:00
fceb00d234
improved comment about global validation throttle
2018-01-14 02:24:13 +02:00
145a84a33b
use a single channel for all validation results
2018-01-13 22:02:28 +02:00
f6081fb061
pushMsg should just call maybePublishMessage when it doesn't need validation
2018-01-13 21:56:57 +02:00
473a5d2873
sendMsg should have a buffer
2018-01-13 21:39:35 +02:00
edcb251ad1
install global validation throttle, use reasonable defaults.
2018-01-13 21:15:40 +02:00
856a25c8eb
WithMaxConcurrency is WithValidatorConcurrency
...
and defaultMaxConcurrency is defaultValidateConcurrency.
2018-01-13 20:56:03 +02:00
bf2151ba5f
the sendMsg channel should yield pointers for consistency
2018-01-13 20:47:28 +02:00
5ef13c764e
don't always spawn a goroutine for sending a new message
2018-01-13 20:11:32 +02:00
c95ed28496
add validation context for cancelation on aborts
2018-01-13 18:34:00 +02:00
fba445bc6d
code cosmetics
...
reword pushMsg for less indentation nesting.
2018-01-13 14:44:33 +02:00
982c4de960
per subscription validation throttle and more efficient dispatch logic
2018-01-13 14:31:34 +02:00
d2f6a0050f
WithValidator and WithValidatorTimeout are subscription options
2018-01-13 12:33:03 +02:00
4241241031
fix dangling maxConcurrency reference
2018-01-13 12:24:31 +02:00
88274db0bb
make maximum concurrency configurable, split loop
2018-01-13 12:15:27 +02:00
fe09d1eea3
make validator timeout configurable
2018-01-13 12:13:54 +02:00
6e8b9f2d5c
fix timeout
2018-01-13 12:13:54 +02:00
02877cda71
complete validator functions
...
- make validators time out after 100ms
- add context param to validator functions
- add type Validator func(context.Context, *Message) bool
- drop message if more than 10 messages are already being validated
2018-01-13 12:13:54 +02:00
197a5982a4
ungxify
2018-01-13 12:12:21 +02:00
7dd4e0bfeb
vet
...
used for range variable inside goroutine, now passed as argument
2018-01-13 12:12:21 +02:00
1945f895a2
log when validator discards message
2018-01-13 12:12:21 +02:00
930f264a27
typedef subscription options and fix typo
2018-01-13 12:12:21 +02:00
647bb98655
optionally allow caller to validate messages
2018-01-13 12:12:21 +02:00
95dce910ba
narrow the message seqno to 64bit again
...
Uses the atomic counter initialized at instantiation time, which saves
a few bytes and a syscall per message.
2018-01-03 20:30:28 +02:00
8a31b41b4c
use atomic counter for message seqnos; timestamp is insufficient
2018-01-03 11:25:20 +02:00
56ea7752e8
better worker cleanup
...
1. Don't hang marking a peer as dead if we're shutting down.
2. No need to "drain" the outgoing channel anymore. This may have been necessary
to prevent a deadlock where the main loop blocked on sending on sending a
message while we waited to tell the main loop that the peer was dead. However,
this is no longer an issue (we never block on sending).
2017-10-14 11:19:21 -07:00
d1d3a55404
Merge pull request #34 from libp2p/fix/go-ipfs/4066
...
drop messages to slow peers
2017-09-04 16:24:19 -07:00
2ac19f2157
Drop announce messages to peers when their buffer is full
2017-09-01 23:10:06 -04:00
63c977c815
Return false when we haven't subscribed to anything
2017-09-01 21:16:41 -04:00
e7faa78d7c
log when dropping messages
2017-08-29 19:42:33 -07:00
02b9c5724d
drop messages to slow peers
...
Instead of spawning a go routine for every message sent to every peer, buffer
them (actually, we already buffer 32 so this didn't need to be changed) and drop
messages when the buffer fills up.
fixes https://github.com/ipfs/go-ipfs/issues/4066
(writing to a channel in a go routine can be dangerous...)
2017-08-29 19:11:58 -07:00
25ef943eea
change from field to bytes to avoid json marshalling corruption
2017-01-09 09:52:07 -08:00
ae48a15d7b
make Subscribe go through SubscribeByTopicDescriptor
2016-11-18 02:23:38 +01:00
1c9a576526
Test, SubscribeByTopicDescriptor and minor improvement
2016-11-18 02:23:38 +01:00
3a30ab4c17
comments
2016-11-18 02:23:38 +01:00
b71e3adbd7
first cleanup wave
2016-11-18 02:23:38 +01:00
822640a482
multiple subscriptions prototype
2016-11-18 02:23:38 +01:00
a8a29fa17e
refactor to use multiple feeds
2016-11-18 02:23:38 +01:00
4dfbdcdca7
add functionality to query which peers have a given topic
2016-10-18 11:13:23 -07:00
Kevin Atkinson
vyzo
Erin Swenson-Healey
Steven Allen
Raghav Gulati
keks
Garrett Thornburg
Jeromy
Jan Winkelmann