|Re: [wtp-dev] Declaring WTP 3.1 M2 build S-3.1M2-20080928142920/|
Helen Zhang wrote:
A junit that fails "randomly" usually indicates that there is a bug somewhere in the code. Either the code itself isn't right, or the unit test itself is incorrect. I think the last green build was about 2 weeks ago, and that was only for an hour. I've made a suggestion on the WTP-Releng list of a way to possibily help narrow this down. Hopefully we can get the builds going faster so that people aren't having to wait 4 hours.The issue is there is that there are random Junit failures every now and then on the build machine that happens when no changes are released or they pass when ran locally, and we've given these a by in the past. But you're right, we should have more successful build and pay more attention to these Junit failures especially since there hasn't been a Junit error free build on the 3.1 stream for a while.
Sounds good. I know from an adopter standpoint, I'm not comfortable using a build that has unit test failures. Along the testing front, people might be interested in the Junit Anti-Patterns:I'll be opening bugs for every Junit failures that I see on the builds in hopes to achieve 0 Junit failing builds again.
Back to the top