18 Jan 2021 |
Passiing | Redacted or Malformed Event | 07:12:18 |
Passiing | Download Image.png | 07:15:33 |
proletarius101 (Old) | _hc: I think the PR for refactoring to based on material components library is almost done. All tests in the CI passed on my machine, except this one: https://proletarius101.gitlab.io/-/fdroidclient/-/jobs/968749468/artifacts/app/build/reports/tests/testFullDebugUnitTest/classes/org.fdroid.fdroid.nearby.LocalHTTPDManagerTest.html#testStartStop | 07:56:31 |
proletarius101 (Old) | Not sure whether it's really a bug | 07:56:38 |
_hc | proletarius101: did you try running it again, e.g. clicking "Retry"? That testhas been a bit flaky, but mostly it works | 08:39:06 |
proletarius101 (Old) | Locally it doesn't pass too. And indeed it occurs at this line: https://gitlab.com/proletarius101/fdroidclient/-/blob/master/app/src/testFull/java/org/fdroid/fdroid/nearby/LocalHTTPDManagerTest.java#L72 | 08:40:55 |
proletarius101 (Old) | Comment it out it will pass | 08:41:08 |
_hc | commenting out that means the test doesn't really run, that's why it would pass | 08:44:09 |
_hc | that job is passing in master https://gitlab.com/fdroid/fdroidclient/-/jobs | 08:47:21 |
proletarius101 (Old) | In reply to @eighthave:matrix.org commenting out that means the test doesn't really run, that's why it would pass Sure. The thing is, what is executed between L55 and L72? I see nothing but tearing down? | 08:54:36 |
_hc | the test is to make sure it can start and stop the internal webserver used in Nearby app swapping | 08:55:36 |
_hc | LocalHTTPDManager.stop(context); | 08:55:44 |
_hc | and LocalHTTPDManager.start(context, false); | 08:55:52 |
_hc | Roboectric and Espresso both seem to require some serious superstition to keep running | 08:56:41 |
_hc | I'm so disappointed in how crappy Android's test setups are, especially considering that Google is sitting on over $50 billion in cash. How about spending a bit of that on making your test suites work well? | 08:57:58 |
_hc | I wonder if Robolectric gets any money from Google at all | 08:58:29 |
| est left the room. | 08:59:43 |
| glennmen left the room. | 09:00:40 |
_hc | hmm, I'll bet I can make some of those tests only run if related code has changed | 09:00:53 |
_hc | or at least that's easy with Python and gitlab-ci | 09:01:15 |
mimi89999 | > I'm so disappointed in how crappy Android's test setups are, especially considering that Google is sitting on over $50 billion in cash. | 09:02:11 |
mimi89999 | Maybe they are building Fuschia? | 09:02:12 |
_hc | ha, no. that's pretty clear | 09:03:11 |
proletarius101 (Old) | In reply to @eighthave:matrix.org Roboectric and Espresso both seem to require some serious superstition to keep running yeah... And there are some test errors that were ignored but after I raise Java version to 8, they appear. And I fixed them. | 09:03:38 |
_hc | I think Kotlin+Fuchsia is basically Google's way of threatening Oracle to not push too hard on the Java lawsuits, or they'll switch to Plan B. You can see that the "native" support stuff is window dressing, they haven't worked on the core GUI stuff that would be required | 09:04:39 |
proletarius101 (Old) | like failure to casting a bool into an Int | 09:04:43 |
proletarius101 (Old) | In reply to @eighthave:matrix.org or at least that's easy with Python and gitlab-ci that would be nice | 09:05:16 |
mimi89999 | And now also Dart | 09:05:32 |
_hc | proletarius101: I apprecate any time you put into troubleshooting that test failure, I think its clearly not related to your changes | 09:07:19 |
mimi89999 | But Fuchsia seems more serious than just a mockup for Oracle. | 09:08:18 |