atais commented 5 years ago

This is what I have done:

PipelineSpec had an assertion that sometimes would fail Some(0) depending on client response. Also, fail checks were missing
Bench had too short timeout, and somtimes failed.
StringOperationsSpectimeout is same as TTL, which would sometimes fail
ListOperationsSpec blocking ops in external Runnable are nasty code. Fixed with Future and blocking futureValue
BlockingDequeSpec is the same case as ListOperationsSpec
WatchSpec - the same as ListOperationsSpec
PatternsSpec did not clean up after tests
PubSubSpec had a test that kept the connection alive, and actually is not a test

About travis:

specifying scala versions spawned multiple machines, however, you can cross-compile using newest scala to older versions. Or the other way round too. Sbt will download the proper version for you.
running sbt with ++{scalaVersion} also ommited 2.10 build which is specified in the sbt. I think it is best to simply let the sbt do the job.
I removed the retry, because the tests should not fail randomly. this needs to be stable.

This can be merged before others, I will handle the conflicts.

debasishg commented 5 years ago

should I merge this one only or the other 2 that precedes this as well ?

atais commented 5 years ago

you can start with this one and the next one is #244

245 I believe is correct but I would like to reassure :) and should go as last

debasishg commented 5 years ago

Will it be better to consolidate all the PRs into 1 and remove the redundant ones ?

atais commented 5 years ago

then please focus on #244 since it contains this one and is my gateway for further improvements :)

merge #244 and close this one

I devided them in logical smaller pieces so it would be easier for you to review.

debasishg / scala-redis