+1 Though I suspect in some areas (particularly in the SQL engine) the answers tend to be much more uniform: unit test + at least some regression test on a fairly uniform workstation environment. (Performance changes are sometimes tested on clusters as well.)
Dave -----Original Message----- From: Carol Pearson [mailto:[email protected]] Sent: Friday, July 22, 2016 9:53 AM To: [email protected] Subject: Proposal - Outline testing in Jira or Pull Request? Hi, It would be very helpful if we could track the environments and cases used to test changes in a particular Jira. Not a gory detail list of 20 test cases and their output, but short and sweet: - Tested positive path - Tested when regionserver is down - Tested with bad arguments for CREATE TABLE - All testing on CDH 5.4.5/Centos 6.x or whatever is applicable. This would be a big help for evaluating pull requests. Plus if your change might behave differently on on different circumstances (different distros or in conjunction with other changes), this can help guide debugging. - guiding "Has this been tried before?" kinds of questions. Thoughts? -Carol P. --------------------------------------------------------------- Email: [email protected] Twitter: @CarolP222 ---------------------------------------------------------------
