Get Travis CI passing consistently #85

ricardopereira · 2015-11-25T08:14:39Z

Fixed:

Domain based on the testing environment.
if algorithm is nil then cipherEncoder should be ignored.
RTC5b.
RSN4: channel as a weak reference but the pointer does not mutate. The object is deallocated but it does not zero out the pointer.

mattheworiordan · 2015-12-02T21:17:55Z

Looks good, would be nice (if available) to have tests marked as pending so that we see this in the test output.

ricardopereira · 2015-12-03T07:38:14Z

Rebase from master.

ricardopereira · 2015-12-03T10:45:17Z

@mattheworiordan Passing.

mattheworiordan · 2015-12-03T11:17:33Z

I change it to pending because it does not work on the test server. I need to investigate more.

Fine, however I would like to know why it isn't passing as well. I think it would make sense, where possible, to fix issues as and when we find them, as opposed to defer all the failing tests till the end of the process. That way the pain is spread out, what do you think?

PS. If you enable a very verbose log level and send us the raw protocol and REST communication, we can confirm if the issue is on our side, and if not, help you to determine what may be wrong on your side

ricardopereira · 2015-12-04T10:49:51Z

Created a script run-tests to be sure we run the same commands on ci server and local machine
It is possible to choose which build tool to perform the tests: env BUILDTOOL
- Options: xctool (default), xcodebuild-pretty, xcodebuild-travis, xcodebuild
- Example: BUILDTOOL=xcodebuild ./Scripts/run-tests.sh
It is possible to run only one test class like: CLASS=ablyTests:ARTRealtimePresenceHistoryTest ./Scripts/run-tests.sh
Changed main scheme to run tests on the same build (the old way was runnning Swift tests and then ObjC tests. Now it runs both on the same execution)
Before testing, does a clean command

mattheworiordan · 2015-12-04T17:12:54Z

@ricardopereira not quite sure what this PR is about anymore, it's titled stats.
Also, the tests are not passing on CI either.

ricardopereira · 2015-12-21T10:55:42Z

Just for the record - some connections remain active after the test:

RealtimeClient__options__should_support_the_same_options_as_the_Rest_client_UsersricardopereiraRepositoriesWhitesmithablyiosablySpecRealtimeClientswift_34]' started.
2015-12-21 07:11:13.533 xctest[14418:4191633] DEBUG: (ARTWebSocketTransport.m:282) 0x7cfc9300 websocket did receive message {"action":8} <---- From another test
2015-12-21 07:11:13.533 xctest[14418:4191633] DEBUG: (ARTWebSocketTransport.m:220) 0x7cfc9300 websocket did disconnect (code 1000) 
2015-12-21 07:11:15.720 xctest[14418:4191584] DEBUG: (ARTRest.m:72) 0x7cef8d40 alloc HTTP
2015-12-21 07:11:15.720 xctest[14418:4191584] DEBUG: (ARTAuth.m:71) setting up auth method Token with key
2015-12-21 07:11:15.721 xctest[14418:4191584] DEBUG: (ARTRest.m:86) initialised 0x7c86a2c0

tcard · 2015-12-24T18:33:10Z

README.md


-Please visit https://support.ably.io/ for access to our knowledgebase and to ask for any assistance.
+To see what has changed in recent versions of Bundler, see the [CHANGELOG](CHANGELOG.md).


tcard · 2015-12-28T10:31:18Z

LGTM, FWIW :)

ricardopereira · 2015-12-31T16:34:47Z

@mattheworiordan I forgot to mention but I finished the remaining work.

tcard · 2016-01-04T10:20:07Z

Sorry to interrupt with this meta-issue, but in every team I've worked with before we tried to keep pull requests focused, atomic and as minimal as possible, to facilitate review, deployment, possible reverts, etc. But I keep seeing new commits in this one, e. g. a1272d9 changing the license, that really depart from this view. In those other teams I would've been asked to close this pull request and make smaller ones, with the commits adequately arranged and squashed for each one. I guess we can spare that to avoid even more distraction, but at least I think we should stop doing it from now on, unless in this team we want to work different to this for some reason.

It is especially hard for me to review this this way, because I'm new and I'm getting used to the codebase and the specs.

ricardopereira · 2016-01-04T10:29:41Z

@tcard This PR is really a mess and it is my fault. It should not be that way. The license has changed on the master but then I did a merge where it was a mistake and now I revert it again. I believe you are struggling with this PR but it will never happen again. I can clarify you anything. I am happy to help.

@mattheworiordan I already reviewed the missing comments.

tcard · 2016-01-04T10:36:24Z

OK, I've seen now the comments about merging master. That was really messy. E. g. we have b5695fc in this PR, then reverted by the merge master, then a1272d9f28a698c6cda06f0092516be12d035a27 doing exactly the same thing.

Can we rebase -i this at least to remove the master merge?

ricardopereira · 2016-01-04T10:41:22Z

@tcard Yes, I tried to revert b78ac4f but there were necessary changes so I chose to recommit the missing parts.

…d: RTC5b. RSN4: channel as a weak reference but the pointer does not mutate. The object is deallocated but it does not zero out the pointer.

…cripts: run-tests.sh and setup.sh - Created a script `run-tests` to be sure we run the same commands on ci server and local machine - It is possible to choose which build tool to perform the tests: env BUILDTOOL Options: xctool (default), xcodebuild-pretty, xcodebuild-travis, xcodebuild Example: BUILDTOOL=xcodebuild ./Scripts/run-tests.sh - It is possible to run only one test class like: `CLASS=ablyTests:ARTRealtimePresenceHistoryTest ./Scripts/run-tests.sh` - Changed main scheme to run tests on the same build (the old way was runnning Swift tests and then ObjC tests. Now it runs both on the same execution) - Before testing, does a `clean` command - Run setup.sh to install tools dependencies

…rror`

…mitIgnoringError`

This reverts commit f5bd8757f655a93ea064418612ebd5391a6e4921.

…lable`

- Added `Aspects`. Allow swizzling where it is possible, without direct access to the code, to replace methods at runtime with custom code.

tcard · 2016-01-04T12:53:55Z

I've done the rebase -i together with @ricardopereira and now it's a bit less messy. However, when rebasing on top of master I had to handle conflicts again and I probably messed up something, so then I had to add 1df3e28. Sorry, it's not ideal, but at this point I think it isn't worth it to clean it up further.

tcard · 2016-01-04T13:05:31Z

(By the way, the changes in the license and README got lost because they probably shouldn't have been there to begin with. They came from commits outside this pull request, like b5695fc, but that also are not on master. They really should be merged to master separately.)

ricardopereira · 2016-01-04T14:41:50Z

@tcard Seems everything fine. I messed up the license & readme with this 9731ecc 😓 I will create a PR with those changes. Thanks again. I learned one more thing about git.

mattheworiordan · 2016-01-04T14:45:05Z

I think we should merge this so that this crazy PR can land and we can return to normality :)

👍

Get Travis CI passing consistently + SwiftWebSocket

ricardopereira mentioned this pull request Nov 25, 2015

ObjC tests: Questionable API usage errors #86

Merged

ricardopereira mentioned this pull request Dec 3, 2015

Priority: Test suite passing #103

Closed

2 tasks

mattheworiordan changed the title ~~Swift test: Stats~~ Swift test: Get Travis CI passing consistently Dec 4, 2015

mattheworiordan changed the title ~~Swift test: Get Travis CI passing consistently~~ Get Travis CI passing consistently Dec 4, 2015

tcard reviewed Dec 24, 2015
View reviewed changes

ricardopereira added 7 commits January 4, 2016 12:40

Fixed URL request. is a Double

40b2b60

Domain based on the testing environment

0232433

Fixed: if algorithm is nil then cipherEncoder should be ignored. Fixe…

b711e3e

…d: RTC5b. RSN4: channel as a weak reference but the pointer does not mutate. The object is deallocated but it does not zero out the pointer.

Fixed: RSC2 (querySyslog returns an AnyGenerator)

e0b35d9

Pending tests

2215c8f

ARTRestTokenTest.testTTLDefaultOneHour: DEBUG=off

930c40e

ricardopereira added 21 commits January 4, 2016 12:51

[objc] Tests: should wait for Channel.attached

b9ee345

SwfitWebSocket: 2.4.0 (removed dev branch)

bbe030e

Added testPresenceLimitIgnoringError and `testHistoryLimitIgnoringE…

906636a

…rror`

Fixed: preconditions are wrong. Should cancel the operation.

c18825e

Podfile: locked versions

b6386bc

Reverted: ARTRest should retain httpExecutor

6b9681a

Tests: timeout of 10 seconds

698b8a6

Reverted: isSyncComplete call is related with channel2

5e4def8

Added comments to testPresenceLimitIgnoringError and `testHistoryLi…

1d24a8a

…mitIgnoringError`

Reverted: removed persistent end date from RestClient.stats

7ce09b7

RTN9a

8b67aea

connection.key

8fd4557

Added ARTConnectionDetails (proxy for tests)

cc936eb

MockTransport.lastReceivedMessage

abfb12c

RSA15a ✅

e3921ab

Reverted: "ARTStatsMessageTypes: avoid exceptions (use optionals)"

253c83d

This reverts commit f5bd8757f655a93ea064418612ebd5391a6e4921.

ARTStats: use non null values with Swift

15834e9

Encoder: encode and decode can return nil, should be marked with `nul…

dfe7ffb

…lable`

Fix #113: decoding stats properties should not return nil

9b44334

Fix TS2: should return zero

bfce505

Fix RSA8a: check if createTokenRequest is called

3529066

- Added `Aspects`. Allow swizzling where it is possible, without direct access to the code, to replace methods at runtime with custom code.

tcard force-pushed the swift-tests-fix-stats branch from b11cd83 to 3529066 Compare January 4, 2016 12:40

A few changes that got lost in a rebase with conflicts.

1df3e28

ricardopereira added a commit that referenced this pull request Jan 5, 2016

Merge pull request #85 from ably/swift-tests-fix-stats

9a600e6

Get Travis CI passing consistently + SwiftWebSocket

ricardopereira merged commit 9a600e6 into master Jan 5, 2016

ricardopereira deleted the swift-tests-fix-stats branch January 5, 2016 13:43

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Get Travis CI passing consistently #85

Get Travis CI passing consistently #85

ricardopereira commented Nov 25, 2015

mattheworiordan commented Dec 2, 2015

ricardopereira commented Dec 3, 2015

ricardopereira commented Dec 3, 2015

mattheworiordan commented Dec 3, 2015

ricardopereira commented Dec 4, 2015

mattheworiordan commented Dec 4, 2015

ricardopereira commented Dec 21, 2015

tcard Dec 24, 2015

tcard commented Dec 28, 2015

ricardopereira commented Dec 31, 2015

tcard commented Jan 4, 2016

ricardopereira commented Jan 4, 2016

tcard commented Jan 4, 2016

ricardopereira commented Jan 4, 2016

tcard commented Jan 4, 2016

tcard commented Jan 4, 2016

ricardopereira commented Jan 4, 2016

mattheworiordan commented Jan 4, 2016


		Please visit https://support.ably.io/ for access to our knowledgebase and to ask for any assistance.
		To see what has changed in recent versions of Bundler, see the [CHANGELOG](CHANGELOG.md).

Get Travis CI passing consistently #85

Get Travis CI passing consistently #85

Conversation

ricardopereira commented Nov 25, 2015

mattheworiordan commented Dec 2, 2015

ricardopereira commented Dec 3, 2015

ricardopereira commented Dec 3, 2015

mattheworiordan commented Dec 3, 2015

ricardopereira commented Dec 4, 2015

mattheworiordan commented Dec 4, 2015

ricardopereira commented Dec 21, 2015

tcard Dec 24, 2015

Choose a reason for hiding this comment

tcard commented Dec 28, 2015

ricardopereira commented Dec 31, 2015

tcard commented Jan 4, 2016

ricardopereira commented Jan 4, 2016

tcard commented Jan 4, 2016

ricardopereira commented Jan 4, 2016

tcard commented Jan 4, 2016

tcard commented Jan 4, 2016

ricardopereira commented Jan 4, 2016

mattheworiordan commented Jan 4, 2016