Websockets transport now creates correct url #268
Conversation
|
@mikaelm12, |
| @@ -37,7 +37,7 @@ public EndToEndTests(ServerFixture serverFixture) | |||
|
|
|||
| [ConditionalFact] | |||
| [OSSkipCondition(OperatingSystems.Windows, WindowsVersions.Win7, WindowsVersions.Win2008R2, SkipReason = "No WebSockets Client for this platform")] | |||
| public async Task WebSocketsTest() | |||
| public async Task BareWebSocketsTest() | |||
There was a problem hiding this comment.
Not sure why the test name was changed...
There was a problem hiding this comment.
Just to be a little more explicit that this test is using bare websockets and not testing the transports. I can change this back
There was a problem hiding this comment.
Yeah, change it back. Other than that it now looks good.
| yield return new object[] { new string('A', 5 * 1024 * 1024 + 32)}; | ||
| yield return new object[] { new string('A', 5 * 1024), new WebSocketsTransport()}; | ||
| yield return new object[] { new string('A', 5 * 1024 * 1024 + 32), new WebSocketsTransport() }; | ||
| yield return new object[] { new string('A', 5 * 1024), new LongPollingTransport(new HttpClient())}; |
There was a problem hiding this comment.
This does not seem right (given the name of the property). If needed at all (and I am not convinced it is) you should create a getter that creates a Cartesian product of what before was MessageSizesData and Transports
| } | ||
| } | ||
|
|
||
| [ConditionalTheory] | ||
| [OSSkipCondition(OperatingSystems.Windows, WindowsVersions.Win7, WindowsVersions.Win2008R2, SkipReason = "No WebSockets Client for this platform")] | ||
| [MemberData(nameof(MessageSizesData))] | ||
| public async Task ConnectionCanSendAndReceiveDifferentMessageSizesWebSocketsTransport(string message) | ||
| public async Task ConnectionCanSendAndReceiveDifferentMessageSizesWebSocketsTransport(string message, ITransport transport) |
There was a problem hiding this comment.
This test (as indicated by its name) is used to test the logic we have in the WebSockets transport to combine messages which don't fit the buffer/have endOfMessage flag set to false. As such this test is specific to WebSocketsTransport and doesn't have to be run against other transports. We already have a test above that test that sending and receiving data using different transports works.
|
Hmm. Travis Failure https://travis-ci.org/aspnet/SignalR/jobs/208377625#L2744 |
|
*sigh* it's something timing out on Travis yet again. I can't figure out why these tests are so flaky only on Travis. It's got to be something about slow machines and task scheduling there, but it's almost impossible to reproduce on any of my normal machines. For now, restarting the build (which I just did) normally works. |
Addresses: #266