meteor-mocha
meteor-mocha copied to clipboard
Why do tests run slower than with `practicalmeteor:mocha`?
Hi guys,
I found that my test-suite runs much slower using this package (SERVER-side ONLY tests) than running them with practicalmeteor:mocha. It is NOT only this package - I have also tried practicalmeteor:mocha-xunit-reporter which is unusable as it hits the 120minute limit on CircleCi.
Here are some results:
dispatch:mocha OSX => 291 passing (26m)
dispatch:mocha CircleCi => NOT yet tested
practicalmeteor:mocha OSX => 263 passing (17m) // WINNER
practicalmeteor:mocha-xunit-reporter OSX => NOT yet tested
practicalmeteor:mocha-xunit-reporter CircleCi => hits the 120minutes limit (=UNUSABLE)
I am just wondering... maybe someone else is experiencing this too. And maybe someone finds a performance hack
@thebarty I'm also perplexed at this. Some of my integration tests failed on timeout when I first started using dispatch. I don't see anything in the dipatcha:mocha code that is blocking. Practicalmeteor:mocha's code is far more complex.
@keyscores Strange stuff, but I can definetly confirm that for my tests this is the correct data. I just had my first successful pass using dispatch:mocha on CircleCi. So the new data looks like this:
dispatch:mocha OSX => 291 passing (26m)
dispatch:mocha CircleCi => 291 passing (24m) // Wohaa! Faster than on OSX. Faster than practicalmeteor:mocha-xunit-reporter
practicalmeteor:mocha OSX => 263 passing (17m)
practicalmeteor:mocha-xunit-reporter OSX => NOT yet tested
practicalmeteor:mocha-xunit-reporter CircleCi => hits the 120minutes limit (=UNUSABLE)
Are you experiencing similar results?
This is the setup I am using (last topic) https://forums.meteor.com/t/enhanced-but-simply-stupid-circleci-setup-that-first-runs-unit-integration-tests-and-then-chimp-acceptance-tests/35086/7
@thebarty Interesting. I don't really have any ideas. Both packages rely on the same core mocha package and operate pretty much the same way for running server tests. We can accept a PR if anyone can figure out some solution.