Hi Everyone, As you may have noticed, we have a regression [1] in the Sorted Merge transform in 2.18.0. We have no meaningful integration tests for Sorted Merge, so this slipped through the cracks.
Sorted Merge is a crucial transform, so this issue will cause a lot of production deployments to fail when they try to upgrade to 2.18.0. Even though we usually do not create patch releases, this regression might be serious enough to break our habit and create a 2.18.1 release. Given the fact that this regression was introduced in a massive burst of XML cleanup changes [2]. It might make sense to wait another week or 2 to see if any other regressions show up. Just to be clear: this is about the fact that there’s a regression. The XML cleanup is tricky but had to be done. This is in no way meant to point fingers; please consider this discussion a blameless post-mortem. My suggested approach would be: 1. Give this discussion 72 hours to allow all developers and PMC members to weigh in. When the discussion is closed, we inform the users that we have found a significant regression in 2.18.0 and will create a patch release. At the same time, we ask them to check for any additional regressions and report asap if they find any. 2. We wait another 2 weeks to see if any other regressions show up, and then create a 2.18.1 patch release before the end of the month (June). 3. We put our focus on creating additional integration tests for transforms and actions in the 2.19.0 release. Let me know what you think. [1] https://github.com/apache/hop/issues/7219 [2] https://github.com/apache/hop/issues?q=is%3Aissue%20state%3Aclosed%20milestone%3A%222.18%22%20cleanup%20xml Regards, Bart
