suggestion: bail on first failing integration test

[lenkan]

This is just a suggestion. I think it makes sense to bail on the test run when we encounter the first failing test. This will give you quicker feedback on the build if a test is failing.

Note that this is only for the integration tests.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (c0727b6) 81.36% compared to head (34e46b3) 81.36%.

Additional details and impacted files

@@             Coverage Diff              @@
##           development     #168   +/-   ##
============================================
  Coverage        81.36%   81.36%           
============================================
  Files               46       46           
  Lines             4149     4149           
  Branches          1030     1030           
============================================
  Hits              3376     3376           
  Misses             741      741           
  Partials            32       32           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[psteniusubi]

I agree with this.
In addition I think we should use for example jest --testTimeout 30000 to specify a fairly long test timeout by default. Then make sure there are no integration tests that "wait forever" for operations or notifications. This because jest reporting of tests failing because of test timeout is quite bad.

[lenkan]

@psteniusubi Good suggestion! Please see #169 for my proposal on the timeout solution.

[psteniusubi]

Is there a config option for --runInBand? I couldn't immediately find it.
Running integration tests in parallel seems to cause issues, so having this enabled by default could be helpful.

[lenkan]

Is there a config option for --runInBand? I couldn't immediately find it. Running integration tests in parallel seems to cause issues, so having this enabled by default could be helpful.

Currently not, but you can work around it: jestjs/jest#3215