aboutsummaryrefslogtreecommitdiffstats
path: root/lib
Commit message (Collapse)AuthorAgeFilesLines
* Fixing subscription callbacksDiego Ballona2015-11-041-3/+6
|
* Merge pull request #69 from mieko/callbacksRafael Mendonça França2015-10-233-32/+27
|\ | | | | | | Use ActiveSupport::Callbacks for Channel subscription callbacks.
| * Use ActiveSupport::Callbacks for Channel subscription callbacks.Mike A. Owens2015-08-243-31/+32
| | | | | | | | | | | | | | | | | | | | | | | | * Rely on AS::Callbacks for callback handling. * Add before_subscribe, after_subscribe, before_unsubscribe and after_unsubscribe convenience methods * alias on_subscribe and on_unsubscribe to after_subscribe and after_unsubscribe, respectively. * Remove `subscribed` and `unsubscribed` from the callback chain: these methods are now executed as the subject of the callbacks. * Update portions of ActionCable to use the more specific callback names.
* | Merge pull request #84 from alexpeattie/nonstandard-logger-fixRafael França2015-10-231-3/+5
|\ \ | | | | | | Fix NoMethodError when using a custom Rails.logger class
| * | Fix NoMethodError when using a custom Rails.logger classAlex Peattie2015-10-061-3/+5
| | |
* | | Make sure the subscription confirmaion is only sent out oncePratik Naik2015-10-191-2/+10
| | |
* | | Freeze the SUBSCRIPTION_CONFIRMATION_INTERNAL_MESSAGE valuePratik Naik2015-10-191-1/+1
| | |
* | | Send subscription confirmation from server to the client to avoid race ↵Pratik Naik2015-10-165-9/+43
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | conditions. Without this, it's very easy to send messages over a subscription even before the redis pubsub has been fully initialized. Now we delay calling the subscription#connected method on the client side until we receive a subscription confirmation message from the server.
* | | Cover stray deps for the logging convenience requireJeremy Daer2015-10-161-0/+4
| | |
* | | Merge branch 'adjust-reconnect'Jeremy Daer2015-10-162-23/+29
|\ \ \
| * | | Tweak reconnect timingJavan Makhmali2015-10-141-14/+15
| | | |
| * | | Remove immediate reconnect to help avoid thundering herd after server restartsJavan Makhmali2015-10-141-4/+0
| | | |
| * | | Avoid returning results of loopJavan Makhmali2015-10-141-0/+1
| | | |
| * | | Add delay before reopeningJavan Makhmali2015-10-141-2/+9
| | | |
| * | | Improve guard against opening multiple web socketsJavan Makhmali2015-10-141-3/+4
| | | |
* | | | Devolve blanket #require to reveal intent and responsibilityJeremy Daer2015-10-1615-71/+96
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Move specific requires close to where they're needed. * Use the private active_support/rails dep to wrap up common needs like eager autoload and module delegation. * Use a single Rails engine rather than an engine and a railtie. * Prefer `AS::JSON.encode` to `Object#to_json`.
* | | | Test against Rails edge by default. CI against 4.2 also.Jeremy Daer2015-10-161-1/+1
| | | | | | | | | | | | | | | | * Don't deep-require to AD::Http::Request since it misses Mime autoload
* | | | Shush some low-hanging Ruby warningsJeremy Daer2015-10-162-2/+2
| | | |
* | | | Websockets -> WebSocket spelling [ci skip]Jeremy Daer2015-10-154-16/+16
| | | |
* | | | Fix the variable name in error message to make sure it does not raise an ↵Pratik Naik2015-10-151-1/+1
| | | | | | | | | | | | | | | | exception
* | | | Subscribe and unsubscribe from the internal redis channels in the primary EM ↵Pratik Naik2015-10-141-2/+2
| | | | | | | | | | | | | | | | thread
* | | | Don't add the current connection to the connections array until after all ↵Pratik Naik2015-10-141-2/+1
| | | | | | | | | | | | | | | | the callbacks are run
* | | | Remove some excessive loggingPratik Naik2015-10-142-3/+0
| | | |
* | | | add railtie and set default allowed_request_origins for developmentLachlan Sylvester2015-10-153-0/+23
|/ / /
* | | Merge pull request #85 from rails/cross-site-forgery-protectionPratik2015-10-122-1/+15
|\ \ \ | | | | | | | | Add support for cross site forgery protection
| * | | Treat ORIGIN as an opaque identifier and do equality comparison with the ↵Pratik Naik2015-10-121-14/+3
| | | | | | | | | | | | | | | | specified whitelist
| * | | Use Array() instead of Array.wrapPratik Naik2015-10-072-2/+1
| | | |
| * | | First take at cross site forgery protectionPratik Naik2015-10-073-1/+27
| |/ /
* | | Make sure active record queries are logged with the current connection tagsPratik Naik2015-10-124-5/+13
| | |
* | | Be sure not to cache an empty string as the connection_identifierPratik Naik2015-10-091-1/+5
| | |
* | | Include request id in statistics to make it to search the logsPratik Naik2015-10-091-1/+6
| | |
* | | Better schedule pubsub subscribePratik Naik2015-10-091-1/+1
| | |
* | | Be sure to initialize @heartbeat_timer. Third time is a charm!Pratik Naik2015-10-081-2/+2
| | |
* | | Setup the heartbeat_timer in next tick to make sure EM reactor loop is runningPratik Naik2015-10-081-2/+4
| | |
* | | Run a single eventmachine timer to send heartbeatsPratik Naik2015-10-085-35/+18
|/ /
* | EventMachine: shush epoll warnings by checking for support before enabling. ↵Jeremy Daer2015-09-021-1/+2
| | | | | | | | Ditto for kqueue.
* | .js.coffee -> .coffeeRafael Mendonça França2015-09-026-0/+0
| | | | | | | | | | | | | | It was initially required, but support for the shorthand has been supported since sprockets 2.1. Eventually 4.x will only support the shorthand version. Just want to get new people using the prefer stuff ASAP.
* | Increment styleJavan Makhmali2015-08-311-3/+2
| |
* | Guard against opening multiple connectionsJavan Makhmali2015-08-311-1/+1
| |
* | Immediately reconnect after first disconnectJavan Makhmali2015-08-281-0/+7
| |
* | Only send "disconnected" notification onceJavan Makhmali2015-08-281-2/+8
| |
* | Simplify WebSocket reconnects and guard against opening multiple connectionsJavan Makhmali2015-08-281-27/+9
| |
* | Merge pull request #65 from ↵Kasper Timm Hansen2015-08-241-1/+1
|\ \ | | | | | | | | | | | | mrlhumphreys/support-identifiers-without-to-global-id Support connection identifiers that don't implement to_global_id
| * | support connection identifiers that don't implement to_global_id by ↵Mark Humphreys2015-08-241-1/+1
| |/ | | | | | | defaulting to to_s
* | Record last 20 Subscription notifications for inspectionJavan Makhmali2015-08-231-1/+12
| |
* | Immediately reconnect when visibilityState changes to "visible"Javan Makhmali2015-08-231-0/+9
|/
* Remove out-of-date AC::Broadcaster referenceKai Chen = 陈恺2015-08-131-2/+0
| | | Remove out-of-data autoload reference of ActionCable::Broadcaster that removed at e1a99a83ca135523ff8513be756f156500999cb8 .
* Clear out the streams when they are stopped. Otherwise we will keep trying ↵Lachlan Sylvester2015-08-111-1/+1
| | | | to stop them.
* Merge pull request #54 from lsylvester/require-to_paramKasper Timm Hansen2015-08-032-1/+2
|\ | | | | Move require of to_param to where it is needed
| * move the require of object/to_param to channel/broadcasting because that is ↵Lachlan Sylvester2015-08-022-1/+2
| | | | | | | | where it is needed.