Clone Tools
  • last updated a few seconds ago
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
Make sure the bus content is valid before processing

  1. … 2 more files in changeset.
Make sure the bus content is valid before processing

  1. … 2 more files in changeset.
ERRAI-979: Provide functionality for CSRF protection

CSRF protection on message bus servlets is enabled by property.

Enabling the property creates a CSRF token on the first POST

request to the server bus.

The token can be written to an HTML page as a JavaScript variable

with a filter, or else the client can acquire it from a challenge

from the server (a 403 response containing the token as a header).

There is also a filter that protects REST endpoints using the same token.

When an Errai REST caller finds the token in a global JavaScript variable,

it will set this as a header for all REST requests.

Errai REST callers will also retry after a challenge from the server

(403 + token in header).

  1. … 18 more files in changeset.
Add missing license headers and update copyright notice in existing headers.

  1. … 3051 more files in changeset.
Revert "Remove old JBoss and Red Hat license headers. Add new "Red Hat and affiliates" header."

Changes from the reverted commit did not preserve original copyright dates.

This reverts commit d8e735dfe2f6ed25fe9e06b753ae2f6154d76b02.

  1. … 3077 more files in changeset.
Remove old JBoss and Red Hat license headers. Add new "Red Hat and affiliates" header.

  1. … 3159 more files in changeset.
Fixed SSE reconnection logic

Fixed SSE reconnection logic

Fixed SSE reconnection logic

Fixed/Introduced backoff logic for SSE

Fixed/Introduced backoff logic for SSE

Fixed/Introduced backoff logic for SSE

Fixed ERRAI-738: CORS support in google chrome 34+

  1. … 1 more file in changeset.
Log status code when websocket is closed

Fixed RPC transmission in not yet fully initialized websocket channel

Setting websocket channel to null after disconnecting

Fixed ERRAI-687: gradual backoff during reconnect attempts

I've introduced a test for this in the 404 error case. Unfortunately, there is logic in the client bus that does different things depending on the specific status code in the response. Testing all HTTP status codes is unfortunately prohibitive: it takes about 20 seconds per test, so we'll have to settle for partial coverage.

The new test also uncovered a problem with HttpPollingTransportHandler: it wasn't actually cleaning up all its timers when the bus told it to stop. This is fixed now too.

The fixes I made to the code itself, then:

* ClientMessageBusImpl.init() will now delay for a growing amount of time when there hasn't been a successful connection attempt in a while

* Fixed leaking timers in HttpPollingTransportHandler that persist after the bus has stopped and apparently thrown the whole transport handler away

It would be nice to randomize the existing throttling mechanism in the transport handlers. It would be even nicer to wire the transport handlers into the bus's own backoff mechanism, since the various handlers seem to operate independently at the moment.

  1. … 1 more file in changeset.
Fixed ERRAI-687: gradual backoff during reconnect attempts

I've introduced a test for this in the 404 error case. Unfortunately, there is logic in the client bus that does different things depending on the specific status code in the response. Testing all HTTP status codes is unfortunately prohibitive: it takes about 20 seconds per test, so we'll have to settle for partial coverage.

The new test also uncovered a problem with HttpPollingTransportHandler: it wasn't actually cleaning up all its timers when the bus told it to stop. This is fixed now too.

The fixes I made to the code itself, then:

* ClientMessageBusImpl.init() will now delay for a growing amount of time when there hasn't been a successful connection attempt in a while

* Fixed leaking timers in HttpPollingTransportHandler that persist after the bus has stopped and apparently thrown the whole transport handler away

It would be nice to randomize the existing throttling mechanism in the transport handlers. It would be even nicer to wire the transport handlers into the bus's own backoff mechanism, since the various handlers seem to operate independently at the moment.

  1. … 1 more file in changeset.
Work in progress on the bus retry backoff fixes + tests

  1. … 8 more files in changeset.
Work in progress on the bus retry backoff fixes + tests

  1. … 8 more files in changeset.
Replace framework calls from LogUtil to gwt-slf4j.

  1. … 18 more files in changeset.
Fixed RPC error reporting so it displays the additionalDetails sent from the server

This was a surprisingly deep-cutting change: RPC error handling used to go through 3

times as many callbacks as it does now (and it ended up displaying a useless client-

side stack trace in the error dialog)

I also eliminated the transportToBusCallback in favour of a new public sendLocal()

method that does what it says. Transports now deliver messages into the bus itself,

which simplifies all message deliveries--not just errors.

  1. … 9 more files in changeset.
Additional fix for ERRAI-646: Avoid resetting SSE channel every 60 seconds

- For SSE, we now send an empty JSON array as data every 45 seconds to avoid inactivity timeouts

- SSE support was removed from StandardAsyncServlet

  1. … 3 more files in changeset.
Additional fix for ERRAI-634: Ensure that long polling handler doesn't schedule a new polling request when there is already one pending

Fixed SSE error detection and reconnect logic. Tested on Firefox and Chrome.

Fixed TransportHandler resource leaks when bus is reinitialized

  1. … 1 more file in changeset.
Fix to compound OT algo (do not invert rule)

  1. … 24 more files in changeset.
engine fixes

  1. … 11 more files in changeset.
Wikipedia Example passes!

  1. … 24 more files in changeset.
Fix to SSE timeout issue.

  1. … 2 more files in changeset.