Github Bot
Poster of Commits
Disable writing settings more directly in tests. (#124)
* Disable writing settings more directly in tests.
We think there may be some settings leaking across tests, which
is... not ideal.
* Disable test parallelism.
This seems like it may be causing some of our problems with flaky
tests.
* Back out previous changes, synchronize instead.
Just experimenting to see if this helps...
* If you scope the reset in the test, it works. We either need to mock KoLCharacter or we need KoLCharacter.reset in the tests.
* Fixed AscensionPath which was setting property garlandUpgrades instedad of gloverPoints
* Fixed AscensionPath which was setting property garlandUpgrades instedad of gloverPoints
* Let's pass Spotless
* Now with no GarlandUpgrades at all.
* changed Init method to protected
* Fix typo, formatting.
qauntum -> quantum, and also run ./gradlew spotlessApply.
* Remove test session file for fakeUserName.
This shouldn't be created under normal circumstances.
* Add global prefs in test/root.
The most notable setting here is fixedThreadPoolSize=1. That might
help with race conditions...
* Revert "Add global prefs in test/root."
This reverts commit d1c8b54, as it
didn't seem to work.
* Add a simple check for test lifetime events.
We're apparently seeing issues saving preferences on Mac. At the very
least, we should make sure that our username is being appropriately
set.
* Change visibility for lifecycle methods.
BeforeAll / AfterAll are not intended to be used with private methods.
(I'm admittedly not sure why these were working before.)
* Wrap a couple more calls to KoLCharacter.reset.
These two were still generating preference files.
* Don't test FightRequest.initializeAfterFight().
This isn't a very interesting aspect of the test, but worse yet, this
interacts with updateCombatData(NOT_IN_A_FIGHT) to generate a request
in a separate thread to log in as the current user.
This is A Bad Thing, in that it probably generates a NPE as a result
of ForbidNetworkAccess, which if uncaught probably causes the JUnit
Platform to be very unhappy with us.
* Track garlandUpgrades AND gloverPoints
Co-authored-by: BadHorseMonkey <58645293+BadHorseMonkey@users.noreply.github.com>
Co-authored-by: Sam Gaus <sam@gaus.co.uk>
View on Github
* Disable writing settings more directly in tests.
We think there may be some settings leaking across tests, which
is... not ideal.
* Disable test parallelism.
This seems like it may be causing some of our problems with flaky
tests.
* Back out previous changes, synchronize instead.
Just experimenting to see if this helps...
* If you scope the reset in the test, it works. We either need to mock KoLCharacter or we need KoLCharacter.reset in the tests.
* Fixed AscensionPath which was setting property garlandUpgrades instedad of gloverPoints
* Fixed AscensionPath which was setting property garlandUpgrades instedad of gloverPoints
* Let's pass Spotless
* Now with no GarlandUpgrades at all.
* changed Init method to protected
* Fix typo, formatting.
qauntum -> quantum, and also run ./gradlew spotlessApply.
* Remove test session file for fakeUserName.
This shouldn't be created under normal circumstances.
* Add global prefs in test/root.
The most notable setting here is fixedThreadPoolSize=1. That might
help with race conditions...
* Revert "Add global prefs in test/root."
This reverts commit d1c8b54, as it
didn't seem to work.
* Add a simple check for test lifetime events.
We're apparently seeing issues saving preferences on Mac. At the very
least, we should make sure that our username is being appropriately
set.
* Change visibility for lifecycle methods.
BeforeAll / AfterAll are not intended to be used with private methods.
(I'm admittedly not sure why these were working before.)
* Wrap a couple more calls to KoLCharacter.reset.
These two were still generating preference files.
* Don't test FightRequest.initializeAfterFight().
This isn't a very interesting aspect of the test, but worse yet, this
interacts with updateCombatData(NOT_IN_A_FIGHT) to generate a request
in a separate thread to log in as the current user.
This is A Bad Thing, in that it probably generates a NPE as a result
of ForbidNetworkAccess, which if uncaught probably causes the JUnit
Platform to be very unhappy with us.
* Track garlandUpgrades AND gloverPoints
Co-authored-by: BadHorseMonkey <58645293+BadHorseMonkey@users.noreply.github.com>
Co-authored-by: Sam Gaus <sam@gaus.co.uk>
View on Github