Hi Yann, That's a good suggestion. Actually we have DSIP-10[1] with detailed plans for UT coverage and the community has been working on it. We also had an email thread[2] discussing related stuff. Giving examples for UT refactoring / improvement and encourage more contributors to participate are covered in our plans as well. At this moment, we are removing `Powermock`[3] from DS project and try to upgrade all test cases from jUnit4 to jUnit5[4], which are the pre-tasks of UT refactoring / improvement. Once these pre-tasks done, we will refactor the UT cases and increase coverage module by module.
You are very welcome to join us in moving DSIP-10 forward! BTW, you could refer to the links below to catch up with the preview discussions : ) [1] https://github.com/apache/dolphinscheduler/issues/10573 [2] https://lists.apache.org/thread/gcf8yfrnt94l1yqpf548cfbmo4o65blo [3] https://github.com/apache/dolphinscheduler/issues/11405 [4] https://github.com/apache/dolphinscheduler/pull/11332 Thanks! *Best Regards,* *Chufeng (Eric) Gao* On Wed, Sep 28, 2022 at 11:12 PM yann ann <[email protected]> wrote: > I suggest creating a dedicated task such as DSIP, which focuses on > increasing the UT coverage value. At the same time, it can attract more new > contributors to join DS. Because UT will be a good PR for first > > On Tue, Sep 27, 2022 at 10:58 PM david zollo <[email protected]> > wrote: > > > Can't agree more. Now it's time for us to update the UT coverage value, I > > recommend setting the value to 60%, How do you think? > > > > > > > > Best Regards > > > > --------------- > > Apache DolphinScheduler PMC Chair & Apache SeaTunnel PPMC > > David > > Linkedin: https://www.linkedin.com/in/davidzollo > > Twitter: @WorkflowEasy <https://twitter.com/WorkflowEasy> > > --------------- > > > > > > On Mon, Sep 26, 2022 at 9:06 PM yann ann <[email protected]> > wrote: > > > > > Dear All, > > > > > > As we discussed in the WeChat group, the current UT coverage of DS is > not > > > high, in order to avoid increasing our historical debt. I suggest if it > > is > > > possible to establish a policy that future PRs must contain the > > > corresponding UT code. Of course, PRs that are not suitable for writing > > UTs > > > are excluded. > > > UT should a threshold condition for new PR. The reviewers need to > > determine > > > whether the current PR is covered by the test case. If not, the > > Contributor > > > should add the related UT, unless this PR does not need UT. > > > Doing so has the following benefits： > > > 1. Reviewers can migrate PRs more quickly, reducing reviewing time. > > > 2. Submitted PR can get basic quality assurance。 > > > 3. No further reduction in current test coverage > > > > > > Sure, of course there may be some "negative impressions". There is a > > small > > > probability that it may reduce someone's enthusiasm for submitting PR. > > But > > > I think the action is generally beneficial and it should be a > consensus. > > > > > > B. R. > > > Yann (GithubID: DarkAssassinator) > > > > > >
